17th International Conference on Electrical Bioimpedance: ICEBI 2019, Joinville, Santa Catarina, Brazil, 9-14 June 2019 (IFMBE Proceedings, Band 71) [1 ed.] 9811334978, 9789811334979

Table of contents :
Preface
Committees
Conference Chair
International Advisory Committee
Organizing Committee
Reviewers List
17th International Conference on Electrical Bioimpedance ICEBI, June 09–14, 2019, Joinville, Brazil
Organizers
In Collaboration with
Sponsored by
Minicourse Abstracts
Low-Cost Applications of BIA, EIS and EIT
Fast Impedance Spectroscopy UsingMinimalistic Hardware
Basics of Electrical Impedancein Human Medicine
Plenary Speakers Abstracts
Miniaturized Organs on a Chipand Bioimpedance
Fast Impedance Spectroscopy UsingMinimalistic Hardware
Modeling and Signal Processing in ImpedanceSpectroscopy: An Overview
Some Electrical Properties of Human Skin
Electrical Impedance Imaging:from Bench to Bedside
Detection of Gram-Positive and Gram-NegativeBacterium by Electrical Bioimpedance
Contents
Bioinstrumentation
Design and Integration of Electrical Bio-Impedance Sensing in a Bipolar Forceps for Soft Tissue Identification: A Feasibility Study
1 Introduction
2 Methods
2.1 EBI Sensing Modeling
2.2 FE Simulation of EBI Sensing with a Bipolar Forceps
2.3 Experimental Evaluation with Ex-Vivo Porcine Tissues
3 Results
4 Discussions
5 Conclusion
References
Influence of Measurement Pattern on RAW-data in Electrical Impedance Tomography
1 Introduction
2 Materials and Methods
2.1 Sensitivity of a Volume Conductor
2.2 Measurement Patterns
2.3 Measurement Hardware
2.4 Water Tank Measurement
3 Evaluation of Water Tank Measurement
4 Discussion and Conclusion
References
Hardware Setup for Tetrapolar Bioimpedance Spectroscopy in Bandages
1 Introduction
2 Methods
3 Results
3.1 Pressure Evaluation
3.2 BIS Evaluation
4 Discussion and Conclusion
References
Selection of Cole Model Bio-Impedance Parameters for the Estimation of the Ageing Evolution of Apples
1 Introduction
2 Materials and Methods
2.1 Bio-Impedance Measurement
2.2 Model Fitting
2.3 Software and Data Analysis
3 Results and Discussion
3.1 Model Fitting
3.2 Electrical Parameters Analysis
4 Conclusions
References
Biosensor Based on Carbon Nanocomposites for Detecting Glucose Concentration in Water
Abstract
1 Introduction
2 Materials and Methods
2.1 Electrode
2.2 Impedance Measurements
3 Results
4 Discussion
5 Conclusion
Acknowledgment
References
Bioimpedance Measurements on Human Neural Stem Cells as a Benchmark for the Development of Smart Mobile Biomedical Applications
Abstract
1 Introduction
2 Materials and Methods
2.1 Development Methodology and Contextual Framework
2.2 Analog and Digital Design
2.3 Embedded Application
2.4 Cell Culture of hVM1
2.5 Carbon Electrode Chips
2.6 Electrical Impedance Spectroscopy
3 Results and Discussion
3.1 Characterization of Cell Proliferation by Electrical Impedance Spectroscopy (EIS)
4 Conclusions
Conflict of Interest
References
Bioimpedance Theory and Modelling
Numerical Simulation of Various Electrode Configurations in Impedance Cardiography to Identify Aortic Dissection
Abstract
1 Introduction
2 Methods
2.1 Simulation Model
3 Results
4 Discussion and Conclusion
Acknowledgment
References
Numerical Simulation of Impedance Cardiogram Changes in Case of Chronic Aortic Dissection
Abstract
1 Introduction
2 Method
3 Simulation Model
4 Results
5 Conclusion
Acknowledgment
References
Analysis of Silicone Additives to Model the Dielectric Properties of Heart Tissue
1 Introduction
2 Methods
2.1 Silicone Samples
2.2 Simulation and Analysis
3 Results
4 Conclusion and Discussion
References
A Short Review of Membrane Models for Cells Electroporation
Abstract
1 Introduction
2 Electroporation Models
2.1 Kinetics Models
2.2 Asymptotic Model
2.3 A Theoretical Study of a Single-Cell Electroporation in a Microchannel
3 Discussion
4 Conclusions
Conflict of Interest
References
Body Composition
Data Views Technology of Bioimpedance Vector Analysis of Human Body Composition
Abstract
1 Introduction
2 Materials and Methods
3 Results and Discussion
4 Conclusion
Conflict of Interest
References
Analysis of Electrical Bioimpedance for the Diagnosis of Sarcopenia and Estimation of Its Prevalence
Abstract
1 Introduction
2 Methodology
2.1 Participants
2.2 Recruitment
2.3 Definition of Sarcopenia and Its Spectrum
2.4 Sociodemographic Characteristics
2.5 Anthropometric Parameters
2.6 Skeletal Muscle Mass Index (SMI)
2.7 Physical Strength
2.8 Physical Performance
2.9 Statistical Analyses
2.10 Statement of Human and Animal Rights and Statement of Informed Consent
3 Results
3.1 Sociodemographic Characteristics
3.2 Anthropometric Parameters
3.3 Skeletal Muscle Mass Index
3.4 Physical Strength
3.5 Physical Performance
3.6 Sarcopenia Prevalence
4 Discussion
5 Conclusion
Conflict of Interest
Sarcopenia in Patients with Chronic Obstructive Pulmonary Disease and Evaluation of Raw Bioelectrical Impedance Analysis Data
Abstract
1 Introduction
2 Methodology
2.1 Lung Function
2.2 Body Composition
2.3 Functional Capacity
2.4 Statistical Analysis
3 Results
4 Discussion
5 Conclusion
Conflict of Interest
Skeletal Muscle Index Using Bioelectrical Impedance for Diagnosis of Sarcopenia in Two Colombian Studies
Abstract
1 Introduction
2 Materials and Methods
2.1 Participants
2.2 Anthropometric Measurements
2.3 BIA Measurements
2.4 Skeletal Mass Estimation (SMI)
2.5 Handgrip Strength
2.6 Statistical Methods
3 Results
4 Discussion and Conclusion
Conflict of Interest
References
Clinical Applications on Bioimpedance
Bioimpedance Measurement to Evaluate Swallowing in a Patient with Amyotrophic Lateral Sclerosis
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussions
5 Conclusions
Conflict of Interest
References
Three Electrode Arrangements for the Use of Contralateral Body Segments as Controls for Electrical Bio-Impedance Measurements in Three Medical Conditions
Abstract
1 Introduction
2 Methods
2.1 Subject
2.2 Equipment
2.3 Electrodes and Electrode Arrangements
2.4 Statistics
3 Results
3.1 Discussion
4 Conclusions
Conflict of Interest
References
Luminal Electrical Resistivity at 50 kHz of the Pig Large Intestinal Wall
Abstract
1 Introduction
2 Methods
2.1 Samples
2.2 Equipment
2.3 Technique
2.4 Statistics
3 Results
4 Discussion
5 Conclusions
Conflict of Interest
References
Evaluating the Effects of Cold Storage on Vascular Grafts Using Bioimpedance Measurement Techniques
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussion
5 Conclusion
Conflict of Interest
References
Tissue Impedance Spectroscopy to Guide Resection of Brain Tumours
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussion
5 Conclusion
References
Electrical Impedance Spectroscopy
Relationships Between Bioimpedance Variables and Gene Expression in Lactuca Sativa Exposed to Cold Weather
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussion and Conclusion
References
Effect of Heating on Dielectric Properties of Hungarian Acacia Honeys
Abstract
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Methods
3 Results
4 Discussion
5 Conclusion
Conflict of Interest
References
Impedance Measurements Sensitive to Complementary DNA Concentrations
Abstract
1 Introduction
2 Methodology
2.1 Experimental Design
2.2 cDNA Amplification of the 16S Ribosomal Subunit
2.3 Impedance Measurements
3 Results
4 Discussion and Conclusion
Conflict of Interest
References
Monitoring Lactobacillus Bulgaricus Growth in Yoghurt by Electrical Impedance
Abstract
1 Introduction
2 Materials and Methods
2.1 Sample Preparation
2.2 Cell Counting
2.3 Electrical Impedance
2.4 Statistical Evaluation
3 Results
3.1 Results of the Microbiological Cell Count Determination During 12 h
3.2 Results of Electrical Impedance Measurement
3.3 Results of Electrical Impedance Based Cell Count Prediction by PLSR Regression
4 Discussion
5 Conclusion
Conflict of Interest
References
Electrical Impedance Tomography
Source Consistency Frequency Difference Electrical Impedance Tomography (sc-fdEIT)
1 Introduction
2 Methods
3 Results
References
A Measure of Prior Information of a Pathology in an EIT Anatomical Atlas
1 Introduction
2 Methodology
2.1 Prior Based on Samples of CT Scans and in vivo Conductivity Measurements
2.2 Numerical Phantom
2.3 Inverse Problem
2.4 Similarity Index
3 Results
4 Discussion
5 Conclusion
References
Functional Segmentation for Electrical Impedance Tomography May Bias the Estimated Center of Ventilation
Abstract
1 Introduction
2 Methodology
3 Results
4 Discussion
5 Conclusion
Acknowledgements
References
Preliminary Results of a Clinical EIM System
Abstract
1 Introduction
2 Methods
3 Results
4 Discussions
4.1 Single-Frequency Form SFI Tomography Results
4.2 Multi-frequency Tissue/Cell Tomography RFI and DFI Results
5 Conclusion
References
Non-linear Bioimpedance Phenomena
Computational Study of Parameters of Needle Electrodes for Electrochemotherapy
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussion
5 Conclusions
Conflict of Interest
References
Other Bioimpedance Application
Bone Fracture Detection by Electrical Bioimpedance: Measurements in Ex-Vivo Mammalian Femur
Abstract
1 Introduction
2 Materials and Methods
2.1 Phantom Construction
2.2 Measurements
3 Results
4 Conclusions and Study Limitations
References
Bioimpedance Technology for Assessing Blood Filling Redistribution in Human Body Regions During Rotation on Short Radius Centrifuge
Abstract
1 Introduction
2 Materials and Methods
3 Results and Their Discussion
4 Conclusions
Conflict of Interest
References
Differences in the Electrical Impedance Spectroscopy Variables Between Right and Left Forearms in Healthy People: A Non Invasive Method to Easy Monitoring Structural Changes in Human Limbs?
Abstract
1 Introduction
2 Materials and Methods
3 Results
4 Discussion and Conclusion
References
Author Index

Citation preview

IFMBE Proceedings Pedro Bertemes-Filho Editor

Volume 72

17th International Conference on Electrical Bioimpedance ICEBI 2019, Joinville, Santa Catarina, Brazil, 9–14 June 2019

IFMBE Proceedings Volume 72

Series Editor Ratko Magjarevic, Faculty of Electrical Engineering and Computing, ZESOI, University of Zagreb, Zagreb, Croatia Associate Editors Piotr Ładyżyński, Warsaw, Poland Fatimah Ibrahim, Department of Biomedical Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur, Malaysia Igor Lackovic, Faculty of Electrical Engineering and Computing, University of Zagreb, Zagreb, Croatia Emilio Sacristan Rock, Mexico DF, Mexico

The IFMBE Proceedings Book Series is an official publication of the International Federation for Medical and Biological Engineering (IFMBE). The series gathers the proceedings of various international conferences, which are either organized or endorsed by the Federation. Books published in this series report on cutting-edge findings and provide an informative survey on the most challenging topics and advances in the fields of medicine, biology, clinical engineering, and biophysics. The series aims at disseminating high quality scientific information, encouraging both basic and applied research, and promoting world-wide collaboration between researchers and practitioners in the field of Medical and Biological Engineering. Topics include, but are not limited to: • • • • • •

Diagnostic Imaging, Image Processing, Biomedical Signal Processing Modeling and Simulation, Biomechanics Biomaterials, Cellular and Tissue Engineering Information and Communication in Medicine, Telemedicine and e-Health Instrumentation and Clinical Engineering Surgery, Minimal Invasive Interventions, Endoscopy and Image Guided Therapy • Audiology, Ophthalmology, Emergency and Dental Medicine Applications • Radiology, Radiation Oncology and Biological Effects of Radiation IFMBE proceedings are indexed by SCOPUS and EI Compendex. They are also submitted for ISI proceedings indexing. Proposals can be submitted by contacting the Springer responsible editor shown on the series webpage (see “Contacts”), or by getting in touch with the series editor Ratko Magjarevic.

More information about this series at http://www.springer.com/series/7403

Pedro Bertemes-Filho Editor

17th International Conference on Electrical Bioimpedance ICEBI 2019, Joinville, Santa Catarina, Brazil, 9–14 June 2019

123

Editor Pedro Bertemes-Filho Department of Electrical Engineering Universidade do Estado de Santa Catarina Joinville, Santa Catarina, Brazil

ISSN 1680-0737 ISSN 1433-9277 (electronic) IFMBE Proceedings ISBN 978-981-13-3497-9 ISBN 978-981-13-3498-6 (eBook) https://doi.org/10.1007/978-981-13-3498-6 © Springer Nature Singapore Pte Ltd. 2020 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

This volume presents the proceedings of the 17th International Conference on Electrical Bioimpedance (ICEBI 2019), which was held in Joinville, Santa Catarina, Brazil, from June 9 to 14, 2019. ICEBI 2019 continued the series of international conferences in the field of electrical bioimpedance with the main goal focused on bringing together scientists and engineers dealing with fundamental and applied research for reporting on the latest theoretical developments and applications in the fields involved. Our organization is confronting a time of many advances in technology, and we are meeting these advances during a time of larger nation-wide and global change. The world of bioimpedance is an exciting area in which to work/study/play, and we will continue to meet and bring inspired people together in forums like this to ensure our research community remains at the cutting edge. The conference covered a wide range of subjects of primary importance for research and development such as Cell Culture; Nonlinear Bioimpedance Phenomena; Vegetable Tissues; Bioimpedance Theory and Modeling; Clinical Applications on Bioimpedance; Organs and Tissues; Skin Impedance; Electrode Modeling; Bioimpedance Instrumentation; Body Composition; Electrical Impedance Tomography; Electrical Impedance Spectroscopy; Magnetic-Electrical Induction Tomography; Magnetoelectric Resonance Tomography; Industrial Impedance Applications; Impedance Biosensors. The papers included in the proceedings reflect the results of multidisciplinary research undertaken by many groups worldwide. Special attention is paid to the development of novel technologies and biosensors, in particular of bio-nanotechnologies and biomaterials for the improvement of sensing specific marks within bioimpedance data. New clinical bioimpedance applications are proposed for use in medicine and biology. Interesting data on novel chemical and biosensors are also reported which are based on nanostructured metal properties. Considerable progress has been achieved at the intersection of nanotechnologies, information technologies, and biomedicine, for example, in health informatics, biomedical signal, and image processing. New theoretical and experimental results are highlighted in such fields as biosensor, DNA information, bioinstrumentation, v

vi

Preface

electrical impedance tomography, cancer diagnosis, electrode modeling, food quality like yogurt and honey, etc. The proceedings reflect the state of the art in controlling the properties of several classes of biological materials for important future applications and medical diagnosis in various fields. It is worth to note that the proceedings include also a number of review papers reflecting the recent achievements in the development of novel models for cells electroporation and guides for resection of brain tumors as well as novel electronic instrumentation devices on their basis. We hope that the papers included in the ICEBI 2019 proceedings will be of interest for established researchers working in multidisciplinary fields of science and technology, young scientists, students, and broad community wishing to get up-to-date information on progress in the fast-developing areas of electrical bioimpedance and biomedical engineering. Pedro Bertemes-Filho Chairman of ICEBI 2019

Committees

Conference Chair Pedro Bertemes Filho

Universidade do Estado de Santa Catarina

International Advisory Committee Uwe Pliquett Orjan G. Martinsen Franco Simini Dejan Križaj Leigh C. Ward Francisco Miguel Vargas Luna Vincent Senez Ryan J. Halter Fernando Seoane Martinez Marco Carminati Pere J. Riu Barry Belmont Tadeusz Palko Seulki Lee Stig Ollmar Carlos Gonzalez Correa Eung Je Woo César A. González

Institute of Bioprocess and Analytical Metrology, Germany University of Oslo, Norway Universidad de la Republica, Uruguay University of Ljubljana, Slovenia The University of Queensland, Australia University of Guanajuato, Mexico University of Lille, France Dartmouth College, USA University of Borås, Sweden Politecnico di Milano, Italy Universitat Politècnica de Catalunya, Spain University of Michigan, USA Warsaw University of Technology, Poland Johns Hopkins University, USA Karolinska Institutet, Sweden Universidad de Caldas, Colombia Kyung Hee University, South Korea Instituto Politécnico Nacional México, Mexico

vii

viii

Committees

Organizing Committee Pedro Bertemes Filho Antonio Heronaldo de Sousa Fabricio Noveletto Uwe Pliquett

Universidade do Estado de Santa Catarina Universidade do Estado de Santa Catarina Universidade do Estado de Santa Catarina Institute of Bioprocess and Analytical Metrology, Germany

Reviewers List Pedro Berteme Filho Julia G. Busarello Wolff John A. Gomez Sanchez Leigh Ward Tobias Menden Alberto Concu Cesar Gonzalez Ørjan Grøttem Martinsen Christian Tronstad Marcio N. Souza Raul G. Lima Tushar Kanti Bera Lucas H. Negri Fabricio Noveletto Volney Coelho Vincenci Olfa Kanoun Tapani Repo Pere J. Riu Franco Simini Vladimir Kolesnikov

Antonio Dell’Osa David Miranda Zhang Tingting Daniela Suzuki Zhao Song Seward Rutkove Uwe Pliquett Vahid Badeli Dejan Krizaj Carlos A. Gonzalez Correa William A. Cruz Castañeda William Lionheart Stig Ollmar Sergey Rudnev Vincent Senez Clara H. Gonzalez-Correa Eung Je Woo Steffen Leonhardt Claudia Marques

17th International Conference on Electrical Bioimpedance ICEBI, June 09–14, 2019, Joinville, Brazil

Organizers Universidade do Estado de Santa Catarina—UDESC Technological Institute of Joinville—FITEJ

In Collaboration with The International Federation for Medical and Biological Engineering (IFMBE)

Sponsored by Universidade do Estado de Santa Catarina—UDESC

Brazilian National Council for Scientific and Technological Development

ix

x

17th International Conference on Electrical Bioimpedance ICEBI

Brazilian Coordination for the Improvement of Higher Education Personnel

Minicourse Abstracts

Low-Cost Applications of BIA, EIS and EIT Franco Simini Universidad de la Republica, Uruguay simini@fing.edu.uy

xiii

Fast Impedance Spectroscopy Using Minimalistic Hardware Uwe Pliquet Institute of Bioprocess and Analytical Metrology, Germany [email protected]

Impedance spectroscopy is a simple and innocuous way for characterizing materials in terms of electrical conductivity and the capability of storing electrical energy depending on frequency. The well-known and mostly practiced way is sweeping through the frequency range of interest and determining the real and imaginary part of the impedance at each frequency. Even with modern chip technology and a variety of IC especially dedicated to single-chip solutions for impedance measurement, complete impedance spectroscopy using resources for just a few dollars is impossible. Today, methods working in the time-domain become increasingly important since they require much less hardware and obtain a complete spectrum within the fraction of the time needed in the frequency domain. A popular approach is the application of a broad bandwidth signal, like, for instance, a multisine wave, while monitoring the answer from the system. This can be the voltage by application of a controlled current or vice versa. In most cases, voltage and current are monitored. This depends on the front-end electronics which will be considered later. The ratio of the Fourier-transformed voltage and current yields the impedance spectrum. In the case of multisine excitation, only the frequencies contained in the applied signal are used for further processing. Although this approach can be realized with a single microcontroller featuring at least a DAC and two ADC channels, it generates a considerable amount of data due to equidistant sampling. For instance, for a frequency range from 10 Hz to 10 MHz, at least 2 million samples should be recorded to avoid undersampling. Moreover, the ADC should be fast enough—at least 20 MS/s. This contradicts the possibility of having minimalistic hardware because cheap IC with two fast ADC and a memory of at least 8 MB (2 channels, 2 byte/sample, 2 MS) do not exist. A much more pronounced reduction in hardware is possible by choosing excitation signals resulting in monotonic system answer – at least for typical biologic materials. Such signals are Dirac step and ramp functions. Most advantageous is the step function because of the simple creation and a broad bandwidth. The typical response of capacitive systems like biologic materials is a sum of exponential functions. Constant phase element yield a distribution of time constants and diffusive effects like electrode

xv

xvi

U. Pliquet

polarization follow square root –functions. The most important feature is the fast relaxation after the step application which slows considerably down within nanoseconds to milliseconds. The majority of data reduction is achieved by fast sampling immediately after the step and slowing down. Finally, not more than 10 samples per decade are necessary. In order to avoid errors due to violation of the sampling theorem, partial integration of the signal is realized by hardware. The AD conversion is done at the end of an integration period and is not time critical because the next AD conversion follows only after the next integration period. For extremely cheap systems, only one sample point per period is taken which extends the measurement time but does not require fast ADC or large memory. Extremely fast systems utilize parallel integrators for the first sample point with multiplexed AD conversion. This avoids the need for fast ADC and allows continuous measurement using a rectangular wave excitation. In principle, the data can be fitted to exponential functions and the resulting time constants and relaxation strengths can be directly used for further processing. Alternatively, each exponential function can be assigned to an RC combination and the impedance spectrum can be easily calculated. By skipping the exponential fit, partially solving the Fourier integral yields the spectrum in frequency domain for voltage and current from where the impedance can be calculated. The analog front-end is critical for the precision of the system and the desired application. It does no matter whether two, three, or tetrapolar systems are used. The analog front-end is independent off the method of electrical characterization.

Basics of Electrical Impedance in Human Medicine Carlos Augusto González-Correa Universidad de Caldas, Colombia [email protected]

Although Electrical Impedance (EI) has been used in the area of Human Medicine for a while, initially in the field of body composition, there is still a long way to go before its more widespread use is reached in this area. Up to today, there is not a specific medical entity in which EI can be considered as the method of first choice. Nevertheless, many applications have been and are being explored. In fact, there is no human body system for which there is not at least one reference in the scientific literature showing results on the use of EBIS. In my opinion, there are different aspects to consider in this situation. The first one may be the fact that more basic science in needed to explain and interpret the information that can be obtained from EBIS data. This is an area where more research from scientist working in the basic biomedical areas is needed. Once this is achieved, simpler ways to parameterize raw data are needed, so that this can be manipulated by end users, who, at the same time, need more straightforward (i.e., simpler to use) devices that give information in terms of health risks (for instance, very low, low, middle, high or very high), presence or absence of a condition (cancer, for instance), prognosis, etc. From my point of view, the EBIS field applied to medicine would benefit a lot of the interactions and cooperation between people working in the three main aspects involved here: the basic sciences (physics, chemistry, mathematics, and engineering), the basic biomedical explanations (i.e., biology, considering both the structural aspects as the functional ones) and the end users (physicians, nurses, physiotherapists, dentists, sport trainers, etc.). As most of the people attending ICEBI are mainly professionals from the basic sciences, the main aim of the curse is to give an overall view of the many possibilities for the applications of EI in the human health and wellbeing areas.

xvii

Plenary Speakers Abstracts

Miniaturized Organs on a Chip and Bioimpedance Vincent Senez LIMMS, CNRS – University of Tokyo, Lille, France [email protected]

Pathogenic microorganisms are widespread in the environment and they can be strong risk factors for specific cancers. Studying host-pathogen interactions with in vitro models is of great importance to accelerate our knowledge about disease mechanisms and design new therapeutic approaches. The advantages of miniaturization technologies are parallelization of tests and integration of sensors with their embedded data analysis. Very sensitive sensing techniques are mandatory since many pathogens can be harmful in very low concentration (i.e., few infective agents). Many quantification methods are well suited for integration. The lookout for Cryptosporidium, one of the most common waterborne parasitic protozoans, outbreaks is a perfect case study to illustrate this technological challenge. With the significant progress in microfabrication technologies, using MEMS to automate the detection process or decrease the detection limit further is now an attractive proposition. Several studies have tried to tackle this problem using various strategies such as impedance spectroscopy, electrochemical biosensor, surface plasmon resonance, quartz microbalance, cantilever-based systems, surface enhanced Raman spectroscopy and flow cytometry reaching sensitivities down to single oocyst/cyst level. Traditional in vivo infectivity assays, performed in neonatal mouse, are costly and time-consuming. We here present our work on in vitro 2D cell layer and organotypic tissue cultures. We also show that an electrical impedance-based device is able to get insights on Cryptosporidium interaction with model of the intestine.

xxi

Fast Impedance Spectroscopy Using Minimalistic Hardware Uwe Pliquet Institut für Bioprozess- und Analysenmesstechnik e.v., Rosenhof 1, 37308 Heilbad Heiligenstadt, Germany [email protected]

Impedance spectroscopy is a simple and innocuous way for characterizing materials in terms of electrical conductivity and the capability of storing electrical energy depending on frequency. The well-known and mostly practiced way is sweeping through the frequency range of interest and determining the real and imaginary part of the impedance at each frequency. Even with modern chip technology and a variety of IC especially dedicated to single chip solutions for impedance measurement, complete impedance spectroscopy using resources for just a few dollar is impossible. Today, methods working in time-domain become increasingly important since they require much less hardware and obtain a complete spectrum within the fraction of the time needed in frequency domain. A popular approach is the application of a broad bandwidth signal, like for instance a multi sine wave, while monitoring the answer from the system. This can be the voltage by application of a controlled current or vice versa. In most cases, voltage and current are monitored. This depends on the front-end electronics which will be considered at later. The ratio of the Fourier-transformed voltage and current yields the impedance spectrum. In case of multisine excitation, only the frequencies contained in the applied signal are used for further processing. Although this approach can be realized with a single microcontroller featuring at least a DAC and two ADC channels, it generates a considerable amount of data due to equidistant sampling. For instance, for a frequency range from 10 Hz to 10 MHz, at least 2 million samples should be recorded to avoid under sampling. Moreover, the ADC should be fast enough at least 20 MS/s. This contradicts the possibility of having minimalistic hardware because cheap IC with two fast ADC and a memory of at least 8 MB (2 channels, 2 byte/sample, 2 MS) does not exist. A much more pronounced reduction in hardware is possible by choosing excitation signals resulting in a monotonic system answer at least for typical biologic materials. Such signals are Dirac step and ramp functions. Most advantageous is the step function because of the simple creation and a broad bandwidth. The typical response of capacitive systems like biologic materials is a sum of exponential functions. Constant phase element yields a distribution of time constants, and diffusive effects, like electrode polarization, follow square root xxiii

xxiv

U. Pliquet

functions. The most important feature is the fast relaxation after the step application which slows considerably down within nanoseconds to milliseconds. The majority of data reduction is achieved by fast sampling immediately after the step and slowing down. Finally, not more than 10 samples per decade are necessary. In order to avoid errors due to violation of the sampling theorem, partial integration of the signal is realized by hardware. The AD conversion is done at the end of an integration period and is not time critical because the next AD conversion follows only after the next integration period. For extremely cheap systems, only one sample point per period is taken which extends the measurement time but does not require fast ADC or large memory. Extremely fast systems utilize parallel integrators for the first sample point with multiplexed AD conversion. This avoids the need for fast ADC and allows continuous measurement using a rectangular wave excitation. In principle, the data can be fitted to exponential functions and the resulting time constants and relaxation strengths can be directly used for further processing. Alternatively, each exponential function can be assigned to an RC combination and the impedance spectrum can be easily calculated. By skipping the exponential fit, partially solving the Fourier integral yields the spectrum in the frequency domain for voltage and current from where the impedance can be calculated. The analog front-end is critical for the precision of the system and the desired application. It does no matter whether two, three, or tetrapolar systems are used. The analog front-end is independent off the method of electrical characterization.

References 1. Grimnes, S., Martinsen, Ø.G.: Bioimpedance and Bioelectricity Basics. Academic Press, Amsterdam (2014) 2. Min, M., Parve, T., Pliquett, U.: Impedance detection. In: Encyclopedia of Microfluidics and Nanofluidics, pp. 25. Springer Verlag (2013). ISBN 978-0-387-32468-5 3. Pliquett, U.: Time-domain based impedance detection. In: Li, J.V., Ferrari, G. (eds.) Capacitance Spectroscopy of Semi-conductors, Chap. 7, pp. 175–214. Pan Stanford Publishing Pte Ltd. (2017)

Modeling and Signal Processing in Impedance Spectroscopy: An Overview Olfa Kanoun Technical University of Chemnitz, Germany [email protected]

Impedance spectroscopy is a measurement method of great importance in medicine, chemistry, and material sciences. The measurement of complex impedance over a wide frequency range opens up possibilities for noninvasive measurements and the measurement of non-accessible quantities. Bioimpedance spectroscopy is reliable, easy to use, safe. It provides information about a biomaterial and its tissue structure in depth. Altogether, combined with the use of low excitation signal levels, bioimpedance spectroscopy is proved to be both efficient and safe to be used for long time intervals to track several physiological changes. In the design of experiments and impedance measurement systems, several aspects should be specifically addressed along with the design of the experiment and the design of the impedance measurement systems, to meet the necessary conditions of impedance spectroscopy as a method, namely linearity, stability, causality, and finiteness. All these conditions need to be carefully investigated and fulfilled to apply the method correctly and avoid several problems in the post-measurement data interpretation stage. Several effects need to be eliminated already within the experiment itself to allow a thorough interpretation of the impedance spectra. Different approaches can be considered for information extraction from impedance spectra, even if not all of them are used equally frequently. Physical and chemical modeling have the advantage of providing a profound insight in a material or a system so that variation change in materials, geometries or structures can be tracked. Other signal processing methods could be used depending on the available calculation resources, complexity level and the scope of available experimental data. For physical modeling, it is necessary to understand the physical and chemical phenomena, taking place within the system. Individual effects can be described by differential equations, partial differential equations, or fractional differential equations elaborated in the fundamental theories and including many unknown parameters as necessary. In real systems, several individual phenomena are taking place at the same time and may overlap in the frequency range. A physical model equation must combine the descriptions of partial phenomena, e.g., in series or parallel. The main difficulties are the model formulation on the one hand and the

xxv

xxvi

O. Kanoun

parameter extraction on the other hand. The model formulation is not always unique, especially if models for individual effects, which themselves have several unknown parameters, need to be joint, the parameter extraction procedure becomes a challenging ill-posed inverse problem [1]. Nonetheless, the main advantage is that the parameters after the extraction can be directly related to the measured variables and can, therefore, be used for its calculation. Simplified mathematical models have the advantage to provide the possibility to realize a suitable model structure for parameter extraction so that unknown parameters can be easily extracted. However, it is not always clear how these parameters follow the phenomena or quantity under investigation. For this reason, once the quantity changes, several parameters change simultaneously. For data analysis, statistical methods [2], machine learning [3], data mining, and deep learning are gaining attention and can extract information for systems even with higher complexity levels. Suitable methods for feature extraction and classification need to be developed and deployed in this case to distinguish the different classes [4]. For this purpose, the training data and the experimental data need to be representative of the real scenario, including all relevant aspects, such as environmental parameters, aging and contamination effects.

References 1. Büschel, P., Tröltzsch, U., Kanoun, O.: Use of stochastic methods for robust parameter extraction from impedance spectra. Electrochim. Acta 56(23), 8069–8077 (2011) 2. Tetuyev, A., Kanoun, O: Method of soil moisture measurement by impedance spectroscopy with soil type recognition for in-situ applications. Tech. Mess. 73(7/8) (2006) 3. Guermazi, M., Kanoun, O., Derbel, N.: Investigation of long time beef and veal meat behaviour by bio-impedance spectroscopy for meat monitoring. IEEE Sens. J. Sensors-9401 (2014) 4. Büschel, P., Tröltzsch, U., Kanoun, O.: Use of stochastic methods for robust parameter extraction from impedance spectra. Electrochim. Acta 56 (2011). Elsevier

Some Electrical Properties of Human Skin Ørjan G. Martinsen Department of Physics, University of Oslo, Norway Department of Clinical and Biomedical Engineering, Oslo University Hospital, Norway [email protected]

The human skin is a complex organ, and for many decades, it has been a popular subject for impedance measurements in a large variety of applications. The measurements can be tailored to focus on different properties of the skin by careful choice of features such as frequency range, electrode system, type of electrodes, and data analysis methods. As an example, wet gel electrodes cannot be used to study the epidermal stratum corneum, since the gel will penetrate into the skin and sweat ducts and greatly modify their electrical properties [1]. Furthermore, a four-electrode system will typically take away a large part of the contribution from the electrode polarization impedance and from the stratum corneum and focus the measurements on deeper tissue [2]. This can also to some extent be achieved by using higher frequencies in a two-electrode system [3]. Measurements of the stratum corneum can, for instance, be used for skin hydration assessment or for measurement of electrodermal response [1, 4]. In another application, measurements at deeper skin layers are used for detecting skin cancer [5]. More recently, it has been revealed that skin has memristive properties at lower frequencies [6]. This is when the applied current influences the skin so that the resistance changes, producing pinched hysteresis curves in the I–V plot. No practical application has so far been based on this finding, but it is likely that skin memristive measurements in the future can lead to new sensor technologies in medicine.

References 1. Tronstad, C., Johnsen, G.K., Grimnes, S., Martinsen, Ø.G.: A study of electrode gels for skin conductance measurements. Physiol. Meas. 31, 1395–1410 (2010) 2. Grimnes S., Martinsen Ø.G.: Sources of error in tetrapolar impedance measurements on biomaterials and other ionic conductors. J. Phys. D: Appl. Phys. 40(1), 9–14 (2007) 3. Martinsen, Ø.G., Grimnes, S., Haug, E.: Measuring depth depends on frequency in electrical skin impedance measurements. Skin Res. Technol. 5, 179–181 (1999) 4. Martinsen, Ø.G., Grimnes, S., Karlsen, J.: Electrical methods for skin moisture assessment. Skin Pharmacol. 8(5), 237–245 (1995)

xxvii

xxviii

Ø. G. Martinsen

5. Aberg, P., Nicander, I., Hansson, J., Geladi, P., Holmgren, U., Ollmar, S. Skin cancer identification using multifrequency electrical impedance – a potential screening tool. IEEE Trans. Biomed. Eng. 51(12), 2097–2102 (2004) 6. Pabst, O., Martinsen, Ø.G., Chua, L.: The non-linear electrical properties of human skin make it a generic memristor. Sci. Rep. 8, 15806 (2018)

Electrical Impedance Imaging: from Bench to Bedside Eung Je Woo Kyung Hee University, South Korea [email protected]

The human body is an electrically conducting object with various ions and charge-carrying molecules in complicated structures of cells, tissues, and organs. Endogenous currents are generated from excitable cells and exogenous currents can be injected or induced by man-made devices. Inside the human body, there exist electric and magnetic field distributions, which are commonly expressed as voltage, current density, and magnetic flux density. In bioelectromagnetism, we study the interplays of these physical quantities related to structure, pathology, function, and metabolism of cells, tissues, and organs. There are numerous research opportunities and challenges when we view bioelectromagnetism as a tool for imaging. In electrical impedance tomography (EIT), electrical currents are injected into the human body and induced voltages are measured on its surface to produce cross-sectional images of internal conductivity and permittivity distributions. Clinical applications of EIT may include lung ventilation monitoring, hemodynamic monitoring, sleep apnea diagnosis, and upper airway imaging. Magnetic resonance electrical impedance tomography (MREIT) provides high-resolution conductivity and current density images by externally injecting currents into the human body during MRI scans. Conductivity tensor imaging (CTI) is an MR-based electrodeless technique for high-resolution conductivity tensor image reconstructions. MREIT and CTI can be used as treatment planning tools for electrical stimulations such as tDCS, DBS, RF ablation, and electroporation. Diagnostic imaging applications may include brain tumors, brain functions, liver cirrhosis, EEG source imaging, and others. After reviewing the basics of electrical impedance imaging, technical details of EIT, MREIT, and CTI are described. Introducing their clinical applications, future studies of these platform technologies are also discussed.

xxix

Detection of Gram-Positive and Gram-Negative Bacterium by Electrical Bioimpedance César A. González Instituto Politécnico Nacional México, Mexico [email protected]

Detecting bacteria in samples and differentiating between gram-negative and gram-positive species is an important challenge. The most common method, gram staining, is very time-consuming. To improve the speed, reliability, and economy of detecting the presence and type of bacteria in samples (e.g., from patients, water treatment, or food processing), scientists have employed many methods, including fluorescence flow cytometry, electrical impedance spectroscopy, and image analysis. In this talk, electrical impedance measurements as an inexpensive and practical technique for real-time detection of bacteria in suspension and for differentiation between two gram-positive (Staphylococcus epidermidis and Bacillus) and two gram-negative (Pseudomonas aeruginosa and Cobetia) species are shown. Hence, volumetric bulk bioimpedance spectroscopy performed on a label-free sample of bacteria in suspension can establish classifiers that can probably distinguish between the presence and absence of bacteria and between bacterial types.

xxxi

Contents

Bioinstrumentation Design and Integration of Electrical Bio-Impedance Sensing in a Bipolar Forceps for Soft Tissue Identification: A Feasibility Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhuoqi Cheng, Diego Dall’Alba, Darwin G. Caldwell, Paolo Fiorini, and Leonardo S. Mattos Influence of Measurement Pattern on RAW-data in Electrical Impedance Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tobias Menden, Tobias Textor, Samantha Schadwinkel, Steffen Leonhardt, and Marian Walter Hardware Setup for Tetrapolar Bioimpedance Spectroscopy in Bandages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stephan Dahlmanns, Alissa Wenzel, Steffen Leonhardt, and Daniel Teichmann Selection of Cole Model Bio-Impedance Parameters for the Estimation of the Ageing Evolution of Apples . . . . . . . . . . . . . . Pietro Ibba, Giuseppe Cantarella, Biresaw Demelash Abera, Luisa Petti, Aniello Falco, and Paolo Lugli Biosensor Based on Carbon Nanocomposites for Detecting Glucose Concentration in Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John Alexander Gomez-Sanchez, Renata Hack, Sergio Henrique Pezzin, and Pedro Bertemes-Filho Bioimpedance Measurements on Human Neural Stem Cells as a Benchmark for the Development of Smart Mobile Biomedical Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . André B. Cunha, Christin Schuelke, Arto Heiskanen, Afia Asif, Yasmin M. Hassan, Stephan S. Keller, Håvard Kalvøy, Alberto Martínez-Serrano, Jenny Emnéus, and Ørjan G. Martinsen

3

11

18

25

33

38

xxxiii

xxxiv

Contents

Bioimpedance Theory and Modelling Numerical Simulation of Various Electrode Configurations in Impedance Cardiography to Identify Aortic Dissection . . . . . . . . . . . Alice Reinbacher-Köstinger, Vahid Badeli, Gian Marco Melito, Christian Magele, and Oszkar Bíró Numerical Simulation of Impedance Cardiogram Changes in Case of Chronic Aortic Dissection . . . . . . . . . . . . . . . . . . . . . . . . . . . Vahid Badeli, Alice Reinbacher-Köstinger, Oszkar Biro, and Christian Magele Analysis of Silicone Additives to Model the Dielectric Properties of Heart Tissue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Leonie Korn, Simon Lyra, Steffen Leonhardt, and Marian Walter A Short Review of Membrane Models for Cells Electroporation . . . . . . Jéssica R. da Silva, Raul Guedert, Guilherme B. Pintarelli, and Daniela O. H. Suzuki

51

55

60 67

Body Composition Data Views Technology of Bioimpedance Vector Analysis of Human Body Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Svetlana P. Shchelykalina, Dmitry V. Nikolaev, Vladimir A. Kolesnikov, Kontantin A. Korostylev, and Olga A. Starunova Analysis of Electrical Bioimpedance for the Diagnosis of Sarcopenia and Estimation of Its Prevalence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clara Helena Gonzalez-Correa, Maria Camila Pineda-Zuluaga, and Luz Elena Sepulveda-Gallego Sarcopenia in Patients with Chronic Obstructive Pulmonary Disease and Evaluation of Raw Bioelectrical Impedance Analysis Data . . . . . . . Maria Camila Pineda-Zuluaga, Clara Helena Gonzalez-Correa, and Luz Elena Sepulveda-Gallego Skeletal Muscle Index Using Bioelectrical Impedance for Diagnosis of Sarcopenia in Two Colombian Studies . . . . . . . . . . . . . . . . . . . . . . . . Clara Helena Gonzalez-Correa, Julio Cesar Caicedo-Eraso, and Diana Rocío Varon-Serna

77

84

92

99

Clinical Applications on Bioimpedance Bioimpedance Measurement to Evaluate Swallowing in a Patient with Amyotrophic Lateral Sclerosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Fu Zhang, Courtney McIlduff, Hilda Gutierrez, Sarah MacKenzie, and Seward Rutkove

Contents

xxxv

Three Electrode Arrangements for the Use of Contralateral Body Segments as Controls for Electrical Bio-Impedance Measurements in Three Medical Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 C. A. Gonzalez-Correa, L. O. Tapasco-Tapasco, and S. Salazar-Gomez Luminal Electrical Resistivity at 50 kHz of the Pig Large Intestinal Wall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 C. A. Gonzalez-Correa, L. O. Tapasco-Tapasco, and S. Ballesteros-Lopez Evaluating the Effects of Cold Storage on Vascular Grafts Using Bioimpedance Measurement Techniques . . . . . . . . . . . . . . . . . . . . . . . . 127 Maryam Amini, Jonny Hisdal, Antonio Rosales, Håvard Kalvøy, and Ørjan Grøttem Martinsen Tissue Impedance Spectroscopy to Guide Resection of Brain Tumours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Mareike Apelt, Gesar Ugen, Levin Häni, Andreas Raabe, Juan Ansó, and Kathleen Seidel Electrical Impedance Spectroscopy Relationships Between Bioimpedance Variables and Gene Expression in Lactuca Sativa Exposed to Cold Weather . . . . . . . . . . . . . . . . . . . . . 141 Diego Albani, Alberto Concu, Lara Perrota, Antonio H. Dell’Osa, Andrea Fois, Andrea Loviseli, and Fernanda Velluzi Effect of Heating on Dielectric Properties of Hungarian Acacia Honeys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 Eszter Vozáry, Zsanett Bodor, Kinga Ignácz, Bíborka Gillay, and Zoltán Kovács Impedance Measurements Sensitive to Complementary DNA Concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 Gerardo Ames, R. Gnaim, J. Sheviryov, A. Goldberg, M. Oziel, E. Sacristán, and César Antonio González Monitoring Lactobacillus Bulgaricus Growth in Yoghurt by Electrical Impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Zsanett Bodor, John-Lewis Zinia Zaukuu, Tímea Kaszab, Anikó Lambert-Meretei, Mahmoud Said Rashed, Zoltan Kovacs, Csilla Mohácsi Farkas, and Eszter Vozáry Electrical Impedance Tomography Source Consistency Frequency Difference Electrical Impedance Tomography (sc-fdEIT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Tingting Zhang, Tong In Oh, and Eung Je Woo

xxxvi

Contents

A Measure of Prior Information of a Pathology in an EIT Anatomical Atlas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Rafael Mikio Nakanishi, Talles Batista Rattis Santos, Marcelo Britto Passos Amato, and Raul G. Lima Functional Segmentation for Electrical Impedance Tomography May Bias the Estimated Center of Ventilation . . . . . . . . . . . . . . . . . . . . 181 Alcendino Jardim-Neto and Juliana Neves Chaves Preliminary Results of a Clinical EIM System . . . . . . . . . . . . . . . . . . . . 186 Wei Wang, Gerald Sze, and Zhao Song Non-linear Bioimpedance Phenomena Computational Study of Parameters of Needle Electrodes for Electrochemotherapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 Jéssica R. da Silva, Raul Guedert, Guilherme B. Pintarelli, and Daniela O. H. Suzuki Other Bioimpedance Application Bone Fracture Detection by Electrical Bioimpedance: Measurements in Ex-Vivo Mammalian Femur . . . . . . . . . . . . . . . . . . . . 203 Antonio H. Dell’Osa, Alberto Concu, Fernando Dobarro, and J. Carmelo Felice Bioimpedance Technology for Assessing Blood Filling Redistribution in Human Body Regions During Rotation on Short Radius Centrifuge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 Svetlana P. Shchelykalina, Milena I. Koloteva, Yulia V. Takhtobina, Yuri I. Smirnov, Alexander V. Smirnov, Galina Yu. Vassilieva, and Dmitry V. Nikolaev Differences in the Electrical Impedance Spectroscopy Variables Between Right and Left Forearms in Healthy People: A Non Invasive Method to Easy Monitoring Structural Changes in Human Limbs? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 A. H. Dell’Osa, A. Concu, M. Gel, A. Fois, Q. Mela, A. Capone, G. Marongiu, A. Loviselli, and F. Velluzzi Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221

Bioinstrumentation

Design and Integration of Electrical Bio-Impedance Sensing in a Bipolar Forceps for Soft Tissue Identification: A Feasibility Study Zhuoqi Cheng1(B) , Diego Dall’Alba2(B) , Darwin G. Caldwell1 , Paolo Fiorini2 , and Leonardo S. Mattos1 1

Department of Advanced Robotics, Istituto Italiano di Tecnologia, Genova, Italy [email protected] 2 Department of Computer Science, University of Verona, Verona, Italy [email protected]

Abstract. This paper presents the integration of electrical bioimpedance sensing technology into a bipolar surgical forceps for soft tissue identification during a robotic assisted procedure. The EBI sensing is done by pressing the forceps on the target tissue with a controlled pressing depth and a controlled jaw opening distance. The impact of these 2 parameters are characterized by finite element simulation. Subsequently, an experiment is conducted with 4 types of ex-vivo tissues including liver, kidney, lung and muscle. The experimental results demonstrate that the proposed EBI sensing method can identify these 4 tissue types with an accuracy higher than 92.82%. Keywords: Electrical bio-impedance · Tissue identification forceps · Electrode configuration · Finite element method

1

· Bipolar

Introduction

Robot-assisted minimal invasive surgery (RMIS) has been increasingly adopted in the clinical setting in the last fifteen years since this technology provides many benefits for both patients and surgeons. From the patient side, RMIS guarantees better outcomes than other more invasive approaches (i.e. open-surgery), similar to other minimally invasive surgical procedures (e.g. laparoscopy) [1]. From the surgeons’ point of view, RMIS guarantees improved dexterity, movements scaling and magnified tri-dimensional video feedback. Unfortunately, RMIS technologies available nowadays suffer from the limitations in terms of sensing modalities. Vision from the endoscopic cameras is the only sensing method for most commercial surgical robots. Due to the intrinsic limitations of endoscopic video images, performing automatic detection of different tissue types based on this data source can not provide the required robustness and reliability for supporting the execution of surgical procedure, especially when the field of view is under poor illumination, partially occluded, or foggy due to surgical smoke [2]. c Springer Nature Singapore Pte Ltd. 2020  P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 3–10, 2020. https://doi.org/10.1007/978-981-13-3498-6_1

4

Z. Cheng et al.

Electrical bio-impedance (EBI) sensing has potential to be very helpful in this application, enabling reliable and robust tissue identification while providing advantages such as low cost and fast detection. For instance, EBI technology is able to identify different tissue types with a needle electrode, as shown in [3]. In [4,5], Cheng et al. exploit the EBI sensor with a concentric electric needle to detect blood during a peripheral intravenous catheterization, showing a significant improvement of the operation success rate. In addition, different systems and techniques based on EBI sensing are proposed and developed for various cancer diagnosis such as hepatic cancer [6], breast cancer [7] and skin cancer [8]. Nevertheless, the capability and accuracy of tissue identification by measuring their EBI property significantly depends on the electrodes’ configuration such as the relative distance and the size of electrodes [9]. Thus, most works in previous literature are based on custom designed EBI sensing probes that avoid this problem by imposing fixed electrode configuration. The adoption of the same solution in RMIS procedures is complex because it would require the design of a novel surgical tool with integrated EBI sensor or a drop-in probe. In this study, we propose to simplify this issue by integrating the EBI sensing capability to existing robotic bipolar forceps. This requires minimum hardware modifications as the electrodes of the bipolar forceps can be readily used as the sensing electrodes. With the proposed system, the surgeon can perform real-time tissue identification by slightly pressing the forceps on the tissue surface, directly during the surgical tissue manipulation.

2 2.1

Methods EBI Sensing Modeling

Figure 1 shows the modeling of EBI sensing using a bipolar forceps. Since this study focuses on soft tissues and the EBI measurement is done by touching the tissue producing a small pressing depth (d ≤ 4 mm), we assume that the induced tissue deformation ensures a complete contact between the electrode surface and the tissue. Electrode 1 and Electrode 2 are the two jaws of the bipolar forceps. To measure the EBI of the tissue, a safe electrical AC voltage (U ) is injected. By obtaining the reciprocal current (I), the electrical impedance (Z) can be calculated as Z = UI . Furthermore, the reciprocal current I can be obtained by integrating the current density J through a cross-sectional area A.   I= J = (−iω εˆE) (1) A

A

Specifically, the current density J at a point p on the area A can be computed as the product of complex permittivity of the contacting tissue (ˆ ε) and the electrical field strength E at that point. Also, the complex permittivity εˆ is a function of tissue conductivity σ, permittivity ε and excitation frequency ω: εˆ = iσ/ω + ε . Moreover, |Z| can be computed as |Z| =

|U |  |ω εˆ|| A E|

(2)

Design & Integration of EBI Sensing for Tissue Identification

5

Fig. 1. Modeling the EBI sensing of soft tissue with a bipolar forceps.

Given that U has a constant amplitude, |Z| depends on two variables: the electric field generated by the jaws of the forceps E and the electric properties of the contacting material εˆ. Assuming the contacting material is homogeneous and a constant frequency ω is used, |ω εˆ| is identical for each tissue type. Thus, we can assume that it is independent of the electric field strength E to simplify the model. → − The electric field strength E is the sum of the electric field generated by − → − → Electrode 1 (E1 ) and Electrode 2 (E2 ), and they can be calculated using the Coulomb’s law:   dq1 dq2 + (3) dE = dE1 + dE2 = 2 2 4πε r 4πε 0 0 r2 S1 S2 1 where ε0 is the electric constant, and S1 and S2 are the contacting area of Electrode 1 and 2 respectively, which are functions of the insertion depth d. ri is the distance from a face element on Electrode i to Point p, and dqi is the charge on the face element. In addition, ri is a function of the electrode  pressing depth d, the jaw opening distance L, and the depth of point h: ri = (h − d)2 + (L/2)2 Consequently, we can find that d and L are the two main parameters that impact the measurement of |Z|. To better characterize their influences, quantitative analysis is done by Finite Element (FE) simulation before testing the proposed EBI sensing method with ex-vivo tissue samples. 2.2

FE Simulation of EBI Sensing with a Bipolar Forceps

As shown in Fig. 2(A), the tissue is modelled as an homogeneous and isotropic gray block (20 × 20 × 20 mm3 ). The two purple prisms represent the electrodes whose designs are derived from the shapes of the surgical forceps’ jaws (Fig. 2(D)). The electrodes are inserted into the tissue with a depth d and a distance L between two electrodes. In the simulation, L was set from 2 to 8 mm with a step of 2 mm, and d was set from 0 to 4 mm with a step of 1 mm. Previous studies indicate that εˆ can increase with tissue compression [10,11]. However, in our case, the tissue samples are thicker (20 mm) and the pressing displacement is small (≤4 mm),

6

Z. Cheng et al.

Fig. 2. (A) The FEM simulation of the EBI sensing with a bipolar forceps. (B) The dimension of the Maryland bipolar forceps. (C) The simulation results of EBI with different d and L.

corresponding to a small compression rate (≤20%). Therefore, the increment of εˆ would be less than 10% during the tissue compression according to the above studies. The change of εˆ is neglected in this study since the impact from the change of E during the pressing of forceps, which will be illustrated later, is much greater. Considering the modeling described in Sect.2.1, we set the conductivity to σ0 S/m and permittivity ε of the tissue to be 0. Therefore, |ω εˆ| is equal to σ0 . This simplification enables the FE simulation to be done using DC voltage, but does not affect the accuracy of the results. The software package ANSYS Multiphysics was used for running the simulation. The results demonstrated that |Z| decreased when d was bigger and L was smaller. The sensitivities of |Z| to d and L were calculated: the sensitivity to d ranged from 449/σ0 Ω/mm to 24.2/σ0 Ω/mm, and the sensitivity to L was from 31.8/σ0 Ω/mm to 6.3/σ0 Ω/mm. The higher sensitivity was found for the smaller value of L and d. Based on the simulation results, we propose to measure |Z| of the touching tissue with a specific jaw opening distance L and a flexible pressing depth d within a range from 2 to 4 mm during the application. This design considers that |Z| is very sensitive to d, and in actual use it is difficult to obtain an accurate d since the measured organ can be moving due to physiological motions. In contrast, since L can be easily controlled by the surgical robotic system and this value has relatively low influence to |Z|, this parameter is controlled to be a specific value. Also, the range of d is set to be from 2 to 4 mm because when the tool tip barely touches the tissue (d = 0 and 1 mm) the tool tips may have

Design & Integration of EBI Sensing for Tissue Identification

7

unstable contact with the tissue, while when the forceps is pressing too deeply (d > 4 mm) complications due to large tissue deformations and the change of tissue electric property can be involved. 2.3

Experimental Evaluation with Ex-Vivo Porcine Tissues

Ex-vivo experiments were conducted to evaluate the proposed sensing method for tissue identification. A prototype of EBI sensor was made for the experiments as shown in Fig. 3(A). The EBI sensor consists of an electrical impedance converter (AD5933, Analog Inc., USA) and a micro-controller (Atmega328P, Atmel Co., USA). It can be directly mounted on top of a daVinci endowrist instruments (Maryland bipolar forceps Ref. 400172), and connected to its proximal end for measuring the EBI of the tissue contacting its jaws using the electrification connections already integrated in the tool. The measured value is sent to a computer via USB. The peak-to-peak voltage of the EBI sensor is set to be 0.4 V in order to satisfy the international standard IEC 60601. In addition, the excitation frequency is set to be 100 kHz, allowing the system to classify most different tissue types according to Kalvoy et al. [3] and Gabriel et al. [12]. Then the EBI sensor was calibrated with several known resistors ranging from 786 Ω to 8.2 kΩ, which covers the range of EBI for most tissue types (please refer to Fig. 4). The error rate was found to be 0.59% in average, and the maximum error rate was found to be 1.2%.

Fig. 3. (A) The prototype of the EBI sensing device; (B) The setup of the ex-vivo experiment.

The experimental setup for measuring the EBI of ex-vivo porcine tissues is shown in Fig. 3(B). Four tissue types were used including muscle, liver, kidney, and lung. The forceps with the EBI sensor was fixed to the 4th stage of a micromotion stage (Siskiyou Co., USA) for controlling its vertical movement. During the experiments, L was fixed to be one of the followings: 2, 4, 6, and 8 mm. The position of the forceps was firstly adjusted to just touch the tissue, and this position was initialized as d = 0. Then we controlled the forceps to move 4 mm downwards, and recorded |Z| in every 1 mm. Five samples were prepared for each tissue type, and 10 measurements were collected for different d and different L.

8

Z. Cheng et al.

Fig. 4. The ex-vivo experimental results of four tissue types with different L and d = 2 to 4 mm.

3

Results

For each tissue type Θi and a specific Lj , the experimental results with d = 2 to 4 mm were grouped as a class. All the classes were found to be normally distributed by the Kolmogorov-Smirnov test (all the p values are >0.05). Subsequently, the maximum likelihood estimation method was used to describe each class as Cij = N (μij , σij ). Figure 4 shows the Gaussian models with a width of 4 standard deviations of the mean in different L. In addition, the confusion matrix based on the ±2σ Gaussian model was calculated and shown in Table 1. Each row of the matrix represents the tissue types determined by the EBI sensing system while each column represents the ground-truth tissue type. The confusion matrix indicates that the four tissue types can be classified with considerably high accuracy (≥92.82%) using the proposed EBI sensing method.

4

Discussions

This study designed and assessed new technology to exploit EBI sensing on bipolar forceps for tissue identification. The confusion matrix for the classification of the four tissue types presented in Table 1 demonstrates that the sensing system can identify these tissue types with high accuracy. In addition, the impact of two acquisition parameters, namely d and L, were investigated in this study. We proposed to measure |Z| with a flexible d in a defined range (2–4 mm). This is because the EBI measurement is very sensitive to the parameter d according to the FE simulation in Sect. 2.2. However, in practice,

Design & Integration of EBI Sensing for Tissue Identification

9

Table 1. The confusion matrix: percentage of correctly classified tissue. L = 2 mm Muscle Liver

Kidney Lung

Muscle Liver Kidney Lung

0 0 96.77% 0.59%

100% 7.18% 0 0

0 92.82% 3.14% 0

0 0 0.09% 99.41%

L = 4 mm Muscle Liver

Kidney Lung

Muscle Liver Kidney Lung

0 0 99.95% 3.32%

97.74% 6.56% 0 0

2.26% 93.44% 0 0

0 0 0.05% 96.68%

L = 6 mm Muscle Liver

Kidney Lung

Muscle Liver Kidney Lung

0 0 97.89% 1.24%

98.03% 6.14% 0 0

1.97% 93.86% 1.14% 0

0 0 0.97% 98.76%

L = 8 mm Muscle Liver

Kidney Lung

Muscle Liver Kidney Lung

0 0 98.69% 4.31%

99.24% 6.77% 0 0

0.76% 93.23% 0 0

0 0 1.31% 95.69%

the pressing depth information is estimated by the kinematic information of the surgical robotic system, and this information may not be very accurate due to the limitation of the robot’s encoder and the movement of the measured organ. In contrast, the proposed EBI measurement protocol requires the parameter L to be known, which can be obtained from the robotic system controller.

5

Conclusion

In this work we have presented an EBI measurement system that could be easily integrated in standard bipolar surgical tools for improving the sensing capabilities during RMIS procedures. The proposed system has been tested on ex-vivo animal tissue samples to evaluate its reliability. We have also identified the acquisition parameters that affect the EBI measurements obtained with a bipolar tool. The results confirm that the proposed system is able to repetitively recognize different types of tissue. Physical simulation has been performed to model the tool-tissue interaction, and the simulation results were confirmed by the experimental ones. Future work will focus on a more accurate FE simulation model involving tissue deformation during forceps pressing on the tissue. In addition,

10

Z. Cheng et al.

the electrical impedance will be measured in multiple frequencies for improving the detection accuracy. Also, the system will change to use current source for signal excitation in order to eliminate issues with contact impedance and guarantee a safer measurement. Acknowledgements. This study has been partially supported from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 742671).

References 1. Pavan, N., Zargar, H., Sanchez-Salas, R., et al.: Robot-assisted versus standard laparoscopy for simple prostatectomy: multicenter comparative outcomes. Urology 91, 104–110 (2016) 2. Moccia, S., Wirkert, S.J., Kenngott, H., et al.: Uncertainty-aware organ classification for surgical data science applications in laparoscopy. IEEE Trans. Biomed. Eng. 65(11), 2649–2659 (2018) 3. Kalvøy, H., Frich, L., Grimnes, S., Martinsen, Ø.G., Hol, P.K., Stubhaug, A.: Impedance-based tissue discrimination for needle guidance. Physiol. Meas. 30, 129 (2009) 4. Cheng, Z., Davies, B.L., Caldwell, D.G., Mattos, L.S.: A new venous entry detection method based on electrical bio-impedance sensing. Ann. Biomed. Eng. 46(10), 1558–1567 (2018) 5. Cheng, Z., Davies, B.L., Caldwell, D.G., Mattos, L.S.: A venipuncture detection system for robot-assisted intravenous catheterization. In: 2016 6th IEEE International Conference on Biomedical Robotics and Biomechatronics (BioRob), pp. 80–86. IEEE (2016) 6. Laufer, S., Ivorra, A., Reuter, V.E., Rubinsky, B., Solomon, S.B.: Electrical impedance characterization of normal and cancerous human hepatic tissue. Physiol. meas. 31, 995 (2010) 7. Zou, Y., Guo, Z.: A review of electrical impedance techniques for breast cancer detection. Med. Eng. Phys. 25, 79–90 (2003) 8. Aberg, P., Nicander, I., Hansson, J., Geladi, P., Holmgren, U., Ollmar, S.: Skin cancer identification using multifrequency electrical impedance-a potential screening tool. IEEE Trans. Biomed. Eng. 51, 2097–2102 (2004) 9. Martinsen, O.G., Grimnes, S.: Bioimpedance and Bioelectricity Basics. Academic press (2011) 10. Moqadam, S.M., Grewal, P., Shokoufi, M., Golnaraghi, F.: Compressiondependency of soft tissue bioimpedance for in-vivo and in-vitro tissue testing. J. Electr. Bioimpedance 6, 22–32 (2015) 11. Dodde, R.E., Bull, J.L., Shih, A.J.: Bioimpedance of soft tissue under compression. Physiol. Meas. 33, 1095 (2012) 12. Gabriel, S., Lau, R.W., Gabriel, C.: The dielectric properties of biological tissues: III. Parametric models for the dielectric spectrum of tissues. Phys. Med. Biol. 41, 2271 (1996)

Influence of Measurement Pattern on RAW-data in Electrical Impedance Tomography Tobias Menden(B) , Tobias Textor, Samantha Schadwinkel, Steffen Leonhardt, and Marian Walter Medical Information Technology, RWTH Aachen University, Aachen, Germany [email protected] https://www.medit.hia.rwth-aachen.de

Abstract. The conductivity distribution inside a volume conductor can be reconstructed with Electrical Impedance Tomography (EIT). Therefore, electrodes on the surface of the volume conductor are used to inject a constant current and measure the resulting surface potentials, which is equivalent to transfer impedances. A sequence of current injections and voltage measurements is called measurement pattern and results in a set of transfer impedances. Various measurement patterns exist and each of them has a specific sensitivity, which influences the distinguishability of an object in the reconstructed image. To compare different patterns we introduced three criteria based on the RAW-measurement and evaluated the performance in a water-tank experiment. Measurement patterns with an increased distance between the injecting and measuring electrodes showed more sensitivity and selectivity in the RAW-data and should be preferably chosen compared to the traditional adjacent pattern. Keywords: Measurement pattern · Sensitivity Eit-system · Electrode configuration

1

· Water-tank ·

Introduction

Electrical Impedance Tomography (EIT) reconstructs static or dynamic images of conductivity distributions into the human body. Therefore, EIT injects a harmless sinusoidal current inside the body and measures the resulting surface potentials, which can be mapped to a conductivity image. The influence of a conductivity change inside the body to the measured voltage is called sensitivity, which is dependent on the position of the electrodes, stimulation current, measurement pattern and the accuracy of the measurement [1]. The distinguishability of a conductivity change is important for the usability of EIT in typical medical applications like lung-monitoring and stroke-detection. The comparison of different measurement strategies has been evaluated mostly in simulative studies, because most EIT-systems have constraints regarding the configurability of injection and measurement channels. For example, c Springer Nature Singapore Pte Ltd. 2020  P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 11–17, 2020. https://doi.org/10.1007/978-981-13-3498-6_2

12

T. Menden et al.

Kleinermann et al. evaluated the singular value decomposition of the Jacobian (sensitivity) matrix of the reconstruction for various measurement pattern [2]. They state, that not the highest data resolution protocol is most important in EIT, rather the maximization of Signal-to-Noise (SNR) ratio is desirable. Moreover, Adler and Kauppinen showed that an increasing distance between the injecting electrodes increases the distinguishability of objects in the middle [1,3]. In a practical experiment from Jossinet et al. [4] a more homogenous sensitivity has been measured for electrodes more distant from each other. To overcome the limited amount of measurement patterns, we used an EIT-system, where the injection and measurement pattern can be freely configured and the number of electrodes is easily scalable. Consequently, we compared the RAW-data-sets of 11 EIT measurement patterns in a water tank experiment.

2 2.1

Materials and Methods Sensitivity of a Volume Conductor

An EIT frame consists of multiple tetra-polar impedance measurement, which uses four electrodes to measure the transfer impedance  1 ¯ ΔZ = JLE · J¯LI dν , (1) ν Δσ where ν describes the whole volume conductor. The lead field of the two injecting and two measuring electrodes is given as J¯LI and J¯LE , respectively [5,6]. The contribution of a conductivity change Δσ to the measured impedance change ΔZ is scaled with the dot product of J¯LI and J¯LE , which will be called sensitivity S. Thus, the sensitivity as well as a conductivity change Δσ has an influence to the measured impedance ΔZ, which is proportional to the measured voltage difference Δu, assuming a tetra-polar impedance measurement with a constant current source. 2.2

Measurement Patterns

In time differential EIT (tdEIT), a set of multiple tetra-polar impedance measurements are used to reconstruct an impedance change between t0 and t1 . Depending on the chosen electrodes for current injection and voltage measurement various sensitivities occur. Because EIT-systems have no infinite SNR-ratio the amount of information about a specific region of the volume differs with the chosen region and measurement pattern, respectively. In theory, the electrodes could be applied anywhere on the body, but out of clinical practicability the electrodes are arranged in a circular way around the thorax for lung monitoring. Due to the circular arrangement a tomographic slice is reconstructed containing information of a lenticular region of the thorax. Traditionally, current is injected through adjacent electrodes and voltage is measured through all remaining electrodes. Then, the injecting electrode pair is rotated and the measurement proceeds on all other remaining electrodes. The so called, adjacent measurement

Inuence of Measurement Pattern on RAW-data in EIT

13

pattern, has been used for a long time, but it turned out, that the sensitivity is high in the border regions and low in the center of the object. Thus, Adler et al. proposed to increase the distance between the electrode pairs [1]. This can be achieved with skip patterns, which leap a electrodes between the current injection and b electrodes between the measurement electrodes, called skip[a, b]. A whole EIT-frame is a set of tetra-polar measurements. The investigated patterns in this work are described in Table 1: Table 1. The number of dependent and in-dependent measurements are given for various measurement pattern with NE = 16 electrodes. Pattern

[a, b] (mindep /mdep ) measurements

Adjacent Drive

[0, 0] (104/208)

Cross Drive

[3, 0] (192/192)

Polar Drive

[7, 0] (96/192)

Pseudo Polar Drive [6,0]

(192/192)

skip-2–2

[2, 2] (104/208)

skip-3–3

[3, 3] (104/208)

skip-6–6

[6, 6] (104/208)

skip-2–6

[2, 6] (192/192)

skip-6–2

[6, 2] (192/192)

skip-3–7

[3, 7] (96/192)

skip-7–3

[7, 3] (96/192)

Due to the reciprocity of current and voltage, measurements with swapped electrodes contain no new information. The number of so called dependent measurements are given by mdep and the number of independent measurements are given by mindep . Half of the measurements are dependent if the number of skipped electrodes is equal for injection and measurement or the skip-distance is N2E . A reduced number of independent measurements reduces the resolution of the EIT-frame. However, a pattern which have a significantly higher SNR compared to the adjacent drive, might compensate the reduced resolution. The traditional way to compare the performance of measurement patterns is the investigation of the Jacobian matrix, which maps a known conductivity distribution onto the resulting voltages. This is also called “forward-solution”. The Jacobian can be conducted from a cylindrical FEM-model and a given measurement pattern. The spectrum of the singular-values of the Jacobian is typically used as a metric for sensitivity and noise performance of the reconstruction [2]. However, this approach is a theoretical one and neglect the influence of the EIT-system. We see additional information in the strength and distribution of the RAW-data of a real measurement setup. Section 2.4 introduces three criteria based on RAW-data to compare the usability of a pattern including the noise level of the EIT-system.

14

2.3

T. Menden et al.

Measurement Hardware

In this paper we used a serial EIT-system similar to the device from Santos et al. [7]. However, we modified the multiplexer module to the effect that each electrode can be freely multiplexed to one of the current channels I+/− or measurement channels M+/− , as depicted in Fig. 1. A further, advantage of such a system is, that the number of electrodes can be extended easily by adding a multiplexer to the measurement bus. I+/-

M+/-

FPGA

1:4 Mux

electrode 1

Control

1:4 Mux

electrode 2

1:4 Mux

electrode 16

pos. 13 pos. 7 pos. 1 nonc sphere

DDS Demod

PGA

Fig. 1. Schematic description of the EIT-system. Each electrode can be freely configured to a current injecting channel I+/− or a measurement channel M+/− . Eit frames were acquired for various configurations at a water-tank with an inhomogeneity at pos. 1 to 13.

2.4

Water Tank Measurement

A water-tank experiment has been used to generate reproducible and realistic data-sets for the proposed patterns. We used a cylindrical tank with a diameter of 28.7 cm, a water level of 38 cm with a conductivity of 17.47 mS cm and constant room temperature. 16 equidistantly steel electrode connectors are placed at a height of 25.2 cm. We assume time differential EIT (tdEIT), which reconstructs changes of impedance over time. Thus, a reference frame uref is compared to a frame umeas acquired at a later point in time. The homogeneous reference frame uref is acquired, while the tank is solely filled with conductive water. For umeas , a non-conductive sphere with diameter 4 cm, as depicted in Fig. 1 is placed at the level of the electrodes on position 1 to 13, respectively. EIT-frames are acquired for each position and configuration at 50 kHz. Ideally, the voltage difference of udif f = umeas − uref contains only the information of the inhomogeneity. However, the whole voltage difference also includes noise of the EIT-system and artifacts, e.g. movement of the water. Therefore, an average of 16 EIT-frames is used in all following calculations. Depending on the electrode configuration different voltage differences occur. To compare the electrode configurations with each other we propose three criteria:  |udif f | · mindep (2) umean−dif f = mdep

Inuence of Measurement Pattern on RAW-data in EIT

15

is the mean voltage difference caused by the sphere. In order to compare the amount of information of a pattern with another, the result is scaled by the number of independent measurements mindep . We take umean−dif f as a value describing how much signal is introduced due to the sphere, called “added signal”. The electrode configuration should maximize umean−dif f . Secondly,  (udif f − udif f )2 ustd−dif f = · mindep (3) mdep

# of measurements

is the standard deviation of the voltage differences and is a measure for the difference in information in all measurements of a pattern. Due to the different number of independent measurements, the criteria is also scaled with mindep . We call this criteria a measure of specificity. A high specificity is desirable, because higher variations around the mean value consequently include more regional information. The last criteria, investigates the distribution of voltage-levels. Therefore, a histogram h is created for udif f . Typically, many measurements of a pattern contribute a small voltage difference and just a few have larger ones. In a histogram, the behavior is similar to an exponential function. We fitted a · eτ x onto the distribution of the histogram, as exemplarily shown in Fig. 2. 80 adajcent measurement at pos. 7 fitted curve

60 40 20 0

0

0.2

0.4

0.6

0.8 udiff

1

1.2

1.4

1.6

1.8

·10

6

Fig. 2. The histogram of the voltage differences shows the distribution of voltage levels for an adjacent measurement on tank-pos. 7.

If a configuration includes more measurements with higher voltage differences, the lower the time constant τ will be. Thus, τ should be preferably low.

3

Evaluation of Water Tank Measurement

EIT-frames were acquired for 13 positions of the inhomogeneity and 11 patterns, respectively. The cylindrical tank is rotationally symmetrical. Therefore, each measurement position 1 to 6 has a reciprocal measurement for position 13 to 8. The measurement results for pos. 8 to 13 were mostly equivalent and was

16

T. Menden et al.

mean of diffs

0.3

0.2

0.1

0 adjacent polar pseudo-polar cross skip-6-6 skip-6-2

std of diffs

0.4

0.2

0 ·104

tau of histogramm

0 −0.5 −1 −1.5 −2

1

2

3

4 Position

5

6

7

Fig. 3. Three performance criteria of EIT RAW-data from a water-tank experiment are shown from top to bottom. umean−dif f gives a measure of “added-signal” to the EIT-frame. ustd−dif f describes the specificity of pattern. The distribution of the measurements of an EIT-frame is characterized by the time constant τ .

therefore not further analyzed in the evaluation. The performance of each pattern is given in Fig. 3. The amount of “added signal” of umean−dif f was the highest for skip-6–6 compared to skip-3–3 and skip-2–2. Therefore, the latter patterns were discarded in the following evaluation. As expected, all patterns show a decrease of umean−dif f from pos. 1 to 7, due to the increased distance to the electrodes. The adjacent pattern has the lowest umean−dif f at all, which is not surprising, where also the ratio of mindep /mdep is the lowest. The behavior of umean−dif f and ustd−dif f is comparable, solely the cross pattern shows an increased selectivity in the border region of the tank and a relatively high decrease in the centered regions. The behavior of skip-2–6 is comparable to skip-6–2 and is therefore not

Inuence of Measurement Pattern on RAW-data in EIT

17

shown in Fig. 3. Both patterns have a favorable high umean−dif f and ustd−dif f . The patterns polar, skip-6-6 and skip-6–2 had a remarkable low time constant τ and showed superior performance.

4

Discussion and Conclusion

The proposed criteria “added signal” of umean−dif f , “specificity” of ustd−dif f and the time constant of the histogram τ have been used to compare the measured EIT-frames of various measurement patterns. All patterns showed a decrease of umean−dif f and ustd−dif f from the border to the middle of the object. Patterns, which uses a skip distance of 6 or higher for the injecting electrodes had a significantly higher umean−dif f , especially in more centered regions. Also the distribution of the voltage-sets of these patterns are shifted to higher values and therefore should provide a higher SNR in a real measurement setup. The investigated influence of a pattern on the RAW-data is especially interesting for the real measurement setup, which includes noise and device constraints. Apart from that, the reconstruction itself provides a sensitivity matrix for each pattern. In a further study, the combination of the real measurement scenario with the sensitivity matrix should be evaluated and might enable the combination of different patterns in the reconstruction of a single frame. Acknowledgements. The authors gratefully acknowledge Financial support provided by the German Research Foundation [Deutsche Forschungsgemeinschaft (DFG), LE 817/20-3].

References 1. Adler, A., Gaggero, P.O., Maimaitijiang, Y.: Adjacent stimulation and measurement patterns considered harmful. Physiol. Meas. 32, 731–744 (2011) 2. Kleinermann, F., Avis, N.J., Judah, S.K., et al.: Image reconstruction using nonadjacent drive configurations (electric impedance tomography). Physiol. Meas. IS 15, A153 (1994) 3. Pasi, K., Jari, H., Jaakko, M.: Sensitivity distribution simulations of impedance tomography electrode combinations. Int. J. Bioelectromag. 7, 344–347 (2005) 4. Jossinet, J., Kardous, G.: Physical study of the sensitivity distribution in multielectrode systems. Clin. Phy. Physiol. Meas. 8, 33 (1987) 5. Geselowitz, D.B.: An application of electrodynamic lead theory to impendance plethysmonography. IEEE Trans. Bio-Med. Eng. 18, 38–41 (1971) 6. Jaakko, M., Plonsey, R.: Bioelectromagnetism: Principles and Applications of Bioelectric and Biomagnetic Fields. Oxford University Press (1995) 7. Aguiar, S.S., Robens, A., Boehm, A., Leonhardt, S., Teichmann, D.: System description and first application of an FPGA-based simultaneous multi-frequency electrical impedance tomography. Sensors (Switzerland) 16(8), E1158 (2016)

Hardware Setup for Tetrapolar Bioimpedance Spectroscopy in Bandages Stephan Dahlmanns1(B) , Alissa Wenzel1 , Steffen Leonhardt1 , and Daniel Teichmann1,2 1

Medical Information Technology, RWTH Aachen University, Aachen, Germany [email protected] 2 Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA https://www.medit.hia.rwth-aachen.de/

Abstract. As the demographic change progresses, medical research begins to focus on geriatric diseases. Our work concentrates on patients who suffer from age-related weakness of connective tissues or dilated venous valves which result in chronic venous insufficiency (CVI). CVI leads to a reduced perfusion of limbs, increased venous pressure and tissue deficiency, especially in the lower leg. As a result, chronic wounds develop that can persist for several decades. In clinical practice, CVI patients with wounds are outpatients who consult a physician for diagnosis every two months. A possible way to improve the interval of diagnosis are monitoring technologies like bioimpedance spectroscopy (BIS), which is capable to detect changes in tissue integrity. Developing a device for BIS in bandages could therefore enable quasi-continuous wound status monitoring and alert the physician if necessary. The presented hardware setup for BIS includes textile based electrodes for tetrapolar measurements that can be integrated into a bandage without reducing the comfort of the patient. Shape and size of the electrodes correspond to those of typical wound dressings. The hardware is based on the device AFE4300 for low energy consumption in place of highly dynamic or continuous measurements, as wound status dynamics are slow. We show that the complex impedance of human tissue can be measured with high precision if the electrodes were covered with compression stockings, as contact pressure enhances electrode-skin response. Keywords: Bioimpedance spectroscopy · Chronic wound application · Textile electrodes · Instrumentation

1

· Clinical

Introduction

Ulcus cruris results from CVI and is the most common type of chronic wound. It is often located at the lower leg and can persist for several decades. This disease comes with high risks of infection, necrosis and increase in wound size. c Springer Nature Singapore Pte Ltd. 2020  P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 18–24, 2020. https://doi.org/10.1007/978-981-13-3498-6_3

Bioimpedance Spectroscopy in Bandages

19

These changes can occur over a few days, nevertheless ulcus cruris patients are generally outpatients and the chronic wound is examined by a dermatologist less than every six weeks. Treatment is limited to the protection of the wound using dressings plus the application of pressure with bandages and compression stockings to reduce the risk of edema. Wound dressings are changed by the patients themselves, relatives or nurses once or twice a week. These non-experts often overlook changes in wound status or edema development [1]. Recent research on chronic wounds includes the monitoring of wound status using camera based approaches, often applied to automatically detect wound size. Studies using BIS as a contact-based monitoring technology for the evaluation of tissue integrity in animals have been published in [2]. There, Swisher et al. showed that edema development and changes in cell wall integrity correlate to reductions in resistance or reactance, respectively. Wound healing monitored in humans shows a significant increase of impedance in long term measurements, with an increase of up to 210% over 40 days for reactance [3]. Our research on the simulation of models for the human lower leg indicates as well that changes in conductivity and permittivity of the subcutaneous fat have an effect on local BIS [4]. If BIS as a low-cost and non-invasive monitoring technology appears suitable for wound status detection, it could be applied to ulcus cruris patients. This would lead to the quasi-continuous evaluation of wound status, including edema development, and therefore improve its treatment significantly.

2

Methods

This paper presents a hardware setup for BIS in bandages1 . The device uses textile electrodes to monitor tissue impedance without reducing the comfort of the patient, which is essential to achieve long term monitoring of outpatients. The setup is based on the analogue front-end AFE4300, uses tetrapolar measurements and is displayed in Fig. 1. The AFE4300 by Texas Instruments (TI) (Dallas, United States) is a low-cost BIS device that generates excitation frequencies between 1 kHz and 255 kHz. Its dynamic range can be specified by applying reference resistances. They were selected to 22 Ω and 330 Ω, as impedance in the wounded and the healthy lower leg has been measured to be in the region of 50 and 100 Ω [3,5]. The current injected into the body is set to 375 μA by the AFE4300 and always lies below 500 μA even with +20% error, so that no impairment of health occurs (DIN EN 60601-1). For evaluation, the setup was controlled with a CC1310 LaunchPad microcontroller. The textile electrodes were designed using the fabric Shieldex Medtex P180 by STATEX (Bremen, Germany), which is silver plated and therefore conductive. It was stretched over a non conductive fabric to ensure sufficient stiffness while 1

In general, ‘bandage’ and ‘dressing’ are both used to describe all fabrics that cover a wound. Here, we will use the term ‘dressing’ to describe the (10 × 10 cm) textile that is in direct contact with the wound. ‘Bandage’ means the fabric that fixates and covers the wound dressing.

20

S. Dahlmanns et al.

maintaining adequate flexibility. Each electrode is 2 cm wide and 10 cm long. As standard wound dressings have a size of 10 × 10 cm, the electrodes could be placed adjacent to the dressings without interfering with the healing process. This placement generates a significant amount of current density through skin and subcutaneous fat, where changes in wound status and edema occur. The connection between the conductive textiles and the PCB was realized using plug-in connectors.

Fig. 1. BIS monitoring device. The electrodes (fabric: Shieldex Medtex P180, dimensions: 2 × 10 cm each) are put inside a pressure cuff to recreate pressures exerted by common compression stockings. The inner electrodes are placed approximately 10 cm apart. The electrodes are connected to a PCB which includes an AFE4300. The setup is controlled using a CC1310 LaunchPad.

Measurements for the validation of the textile electrode setup were performed at the lower leg of a healthy volunteer (first author, male, 72 kg, age 28). The electrodes were integrated into a pressure cuff and their surface moisturized. Two measurement sets were performed on two consecutive days to compare changes in impedance due to contact pressures and bandage replacement with those due to long term effects like hydration, sweat or other natural causes. On the first day, pressure evaluation was carried out applying an excitation frequency of 32 kHz. On the second day, BIS was executed with excitation frequencies between 8 kHz an 128 kHz, where β-dispersion is detectable [6], and bandage replacement was simulated. Each measurement was performed multiple times to analyze the precision of the device. Between each measurement, the AFE4300 was automatically reset and re-calibrated to warrant independent measurements.

Bioimpedance Spectroscopy in Bandages

21

The pressures applied for the first evaluation are given in mmHg in Table 1 and correspond to those of compression stockings used in CVI therapy, whose pressure values are given in kPa. Table 1. Pressures applied for the evaluation of electrode-skin response Compression classes

Pressure value

Selected pressure

No compression

0.05), which is typically used to describe the flow of current in the intracellular fluids. This is due to the behaviour of the CPE in the equivalent circuit (1A), that depicts the cell membrane and acts differently depending on the applied frequency. In fact, at high frequency Rp is short-circuited by the CPE, and the impedance behaviour is mainly defined by Rs . On the other hand, at low frequency the CPE acts as an open circuit, forcing the flow of current in the extracellular fluids, resulting in an impedance output

30

P. Ibba et al.

Table 1. Effect of the ageing of apples on the electrical parameters of the fitted equivalent circuit (Cole model). Values represents the mean ± SD of 9 fruits at each day of ageing. Mean values with no common letters are statistically different according to LSD (p ≤ 0.05). Series resistance; Rs (Ω)

Parallel resistance; Rp (Ω) ab 39001, 3 ± 11511, 9 a

CPE-T; Q (Ω−1 )

CPE-P; n (a. u.)

1, 75e − 08 ± 5, 28e − 09 a

0, 750 ± 0, 0243 a

2, 09e − 08 ± 9, 44e − 09 a

0, 738 ± 0, 0361 ab

Day 2

988, 59 ± 82, 40

Day 4

1023, 32 ± 121, 0 a

Day 6

984, 72 ± 140, 7

ab 31455, 4 ± 8917, 9

ab 3, 04e − 08 ± 2, 03e − 08 ab 0, 717 ± 0, 0487 abc

Day 8

921, 58 ± 132, 2

ab 27636, 2 ± 8347, 3

bc 3, 75e − 08 ± 2, 51e − 08 ab 0, 704 ± 0, 0498 bc

37413, 4 ± 11374, 6 a

Day 10 900, 51 ± 98, 81

b

Day 12 915, 73 ± 101, 6

ab 21007, 9 ± 8104, 4

p value 0, 156

23031, 1 ± 7649, 7 0, 0004

bc 4, 47e − 08 ± 2, 89e − 08 b

0, 693 ± 0, 0495 c

c

4, 64e − 08 ± 3, 17e − 08 b

0, 692 ± 0, 0527 c

0, 034

0, 0381

represented by the series of Rs and Rp . The other extracted parameters (Rp , CPE-T and CPE-P), show statistically significant differences among the considered days (p ≤ 0.05), especially between the first and the last days of acquisition. These parameters are respectively related to the extracellular resistance (Rp ), to the cell membrane pseudo-capacitance (CPE-T) and to the heterogeneity of size and shape of the fruit cells (CPE-P). Figure 3 depicts the distribution and the evolution of the values of the circuit parameters couples Rp - CPE-T (Fig. 3A) and Rp - CPE-P (Fig. 3B), for the 9 apples in day 4, 8 and 12 of ageing. These data points were selected according to the preliminary results of the impedance curve trends (Fig. 2) and of the ANOVA (Table 1), that highlighted no significant differences between adjacent stages, symbolized by a common letter between two consecutive mean values in Table 1.

Fig. 3. Dispersion diagram of the distribution of the three selected ageing stage (corresponding to Day 4, 8 and 12) according to Rp and CPE-T (A) and Rp and CPE-P (B). All data (Rp , CPE-T and CPE-P) are normalized by their value at Day 1. Each point depicts a single apple at the given ageing stage.

Selection of Cole Model Bio-Impedance Parameters

31

The main factor contributing to the fruit ageing appears to be the parallel resistance, which evolves from around 37 kΩ in Day 4 to around 21 kΩ in the last day of acquisition (see Table 1), clearly contributing in the clustering of the different ageing stages. However, by means of the Rp only, the difference between each studied fruit cannot be highlighted. These differences are appreciated, especially in the last days of acquisition, by the analysis of the CPE parameters evolution, which represents a change in tissue homogeneity and cell membrane capacitance, depicting a different rate of textural degradation among the nine considered apples. The above-mentioned results contribute to give a better insight of the processes occurring during the fruit ageing. Indeed, the selection of the equivalent circuit parameters helps in the simplification of the correlation of the impedance measurement with the quality attributes of the fruit, and with the discrimination of different physiological conditions. However, there is still a strong need for an evolution of this technique. In this regard, further analysis should be devoted in understanding the influence of the electrode-fruit interface in the electrical measurement. Additionally, the correlation of the impedance with fruit quality parameters (such as sugars, acids, texture and ripening stage) and the development of predictive models, can pave the way for a fast and reliable method for the evaluation of fruit ageing for commercial purposes.

4

Conclusions

In this paper, a microcontroller based EIS system was used to extract the impedance data of apples and to monitor the progress of their ripening during 12 days of storage at room temperature in the 100 Hz–85 kHz frequency range. The resulting bio-impedance output was fitted with the Cole equivalent circuit, extracting the four variables of the circuit (Rs , Rp , and CPE magnitude and phase) and evaluating them in terms of their impact on the fruit ageing evolution. The results demonstrated how the parallel resistance (Rp ) evolution can efficiently represent the general ageing trend of the studied fruit, with discrete variation from the first day to the last day of acquisition. Furthermore, the CPE parameters effectively allow separating the single fruits within the single days of acquisition, resulting useful for a possible texture-wise discrimination of apples. Acknowledgement. This work has been partially supported by the Italian Institute of Technology (IIT). Conflict of Interest. The authors declare that they have no conflict of interest.

32

P. Ibba et al.

References 1. Mu˜ noz-Huerta, R.F., Ortiz-Melendez, A., Guevara-Gonzalez, R.G., et al.: An analysis of electrical impedance measurements applied for plant N status estimation in lettuce (Lactuca sativa). Sensors (Switzerland) 14, 11492–11503 (2014). https:// doi.org/10.3390/s140711492 2. Grossi, M., Ricc` o, B.: Electrical impedance spectroscopy (EIS) for biological analysis and food characterization: A review. J. Sens. Sens. Syst. 6, 303–325 (2017). https://doi.org/10.5194/jsss-6-303-2017 3. El Khaled, D., Castellano, N.N., Manzano-Agugliaro, F., et al.: Cleaner quality control system using bioimpedance methods: a review for fruits and vegetables. J. Clean. Prod. 140, 1749–1762 (2017). https://doi.org/10.1016/j.jclepro.2015.10. 096 4. Chowdhury, A., Bera Tushar, K.: Design and development of microcontroller based instrumentation for studying complex bioelectrical impedance of fruits using electrical impedance spectroscopy. J. Food Process. Eng. 41, e12640 (2017). https:// doi.org/10.1111/jfpe.12640 5. Ibba P, Falco A, Rivadeneyra A, Lugli P.: Low-cost bio-impedance analysis system for the evaluation of fruit ripeness. In: IEEE SENSORS Proceedings, IEEE Sensors 2018, New Dehli, India, pp. 1–4 2018. https://doi.org/10.1109/ICSENS.2018. 8589541 6. Martinsen, O., Grimnes, S.: Bioimpedance and Bioelectricity Basics. Academic Press, Cambridge (2014) 7. Cole K.S.: Permeability and impermeability of cell membranes for ions. In: Cold Spring Harbor Symposium on Quantitative Biology, vol. 8, pp. 110–122 (1940). https://doi.org/10.1101/sqb.1940.008.01.013 8. Freeborn, T.J.: A survey of fractional-order circuit models for biology and biomedicine. IEEE J. Emerg. Sel. Top. Circ. Syst. 3, 416–424 (2013). https:// doi.org/10.1109/JETCAS.2013.2265797 9. Gonz´ alez-Araiza, J.R., Ortiz-S´ anchez, M.C., Vargas-Luna, F.M., Cabrera-Sixto, J.M.: Application of electrical bio-impedance for the evaluation of strawberry ripeness. Int. J. Food Prop. 20, 41044–1050 (2017). https://doi.org/10.1080/ 10942912.2016.1199033 10. Analog Devices: 1 MSPS, 12-Bit Impedance Converter, Network Analyzer. AD5933 datasheet, September 2005. Accessed April 2017 11. Montgomery, D.C.: Design and Analysis of Experiments. Wiley, New York (2001) 12. Yovcheva, T., Voz´ ary, E., Bodurov, I., Viraneva, A., Marudova, M., Exner, G.: Investigation of apples’ aging by electric impedance spectroscopy. Bul. Chem. Commun. 45, 68–72 (2013) 13. Watanabe, T., Ando, Y., Orikasa, T., Kasai, S., Shiina, T.: Electrical impedance estimation for apple fruit tissues during storage using Cole-Cole plots. J. Food Eng. 221, 29–34 (2018). https://doi.org/10.1016/J.JFOODENG.2017.09.028 14. Fang, Q., Liu, X., Cosic, I.: Bioimpedance study on four apple varieties. In: IFMBE Proceedings of 13th International Conference on Electrical Bioimpedance, Graz, Austria, vol. 17, pp. 114–117 (2007). https://doi.org/10.1007/978-3-540-738411 32 15. Bakr, A.A., Radwan, A.G., Madian, A.H., Elwakil, A.S: Aging Effect on apples bioimpedance using AD5933. In: ACTEA Proceedings of 3rd International Conference on Advances in Computational Tools for Engineering Applications, Zouk Mosbeh, Lebanon, pp. 158–161 (2016). https://doi.org/10.1109/ACTEA.2016.7560131

Biosensor Based on Carbon Nanocomposites for Detecting Glucose Concentration in Water John Alexander Gomez-Sanchez1,2(&), Renata Hack1, Sergio Henrique Pezzin1, and Pedro Bertemes-Filho1,2 1

Universidade do Estado de Santa Catarina, UDESC, Florianópolis, Brazil [email protected] 2 Department of Electrical Engineering, Florianópolis, Brazil

Abstract. New materials have been developed with nanotechnology since the 1960s. Carbon-based nanocomposites are used as biosensor due to structural, electrical and thermal properties. These nanocomposites have advantages as high sensitivity, mechanical flexibility, and biocompatibility useful in glucose sensors. Non-conventional methods have been used to measure glucose in physiological fluids like urine, sweat, saliva, breath, interstitial and ocular fluid. This work presents preliminary results of electrical impedance spectroscopy measurements at low glucose concentrations (36–1000 µM). It uses bipolar sensor coated with a mixture of DBEGA and graphene. Results indicate a higher sensitivity (348 Ω/µM) in glucose concentrations lower than 200 µM. The sensitivity diminished with increments of glucose concentration. This sensor has a promising application to quantify glucose concentrations in frequencies between 100 Hz to 10 MHz with low Electrode-Electrolyte Interface. A possible sensor application is monitoring continuously physiological aqueous media as urine, sweat or saliva sensing for hypoglycemic and hyperglycemic patients Keywords: Biosensor


Graphene
Glucose

1 Introduction Carbon based nanocomposites as carbon black, carbon nanotubes, fullerenes, nanofibers and graphene has been used in medical applications as biosensors, Graphene has a hexagonal geometry with a planar network structure in a monolayer of sp2 carbon atoms [1]. Graphene is a semi-conductor with higher carrier’s mobility containing oxygenated functional groups; this compose shows several advantages compared with another material, some characteristics are: high sensitivity, suitability of miniaturization, mechanical flexibility and biocompatibility [2]. These characteristics has been used in glucose sensors due higher electrocatalytic in 02-saturate condition [3]. Carbon nanocomposites sensors are good alternative due to their low-cost, good time-response, simplicity construction, and easy measurements [4]. Non-conventional methods to measure glucose in alter-native physiological fluids has been found in literature, such as interstitial fluid, urine, sweat, saliva, breath and ocular fluid [5]. Urine glucose levels are used as an indicator for diabetes when glucose is poorly controlled. Glycemic disorders are higher associated with urinary infections in diabetes patient [6]. Glucose © Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 33–37, 2020. https://doi.org/10.1007/978-981-13-3498-6_5

34

J. A. Gomez-Sanchez et al.

sensor based on sweat measurements are used with wearable sensor sets connected to a portable electrochemical analyzer. The sensor absorbs sweat to detect glucose and pH improving accuracy in glucose measurements, the records are correlated using a commercial glucose device [7]. Correlation between human saliva samples with glucometer measurements are also found using non-enzymatic sensor [8]. A sensor with higher sensitivity for lower glucose concentrations is an advantage in measurements of physiological fluids as sweat and saliva. We propose a sensor to detect lower concentrations of glucose in aqueous physiological solution using a bipolar sensor coated with a mixture of DBEGA and graphene.

2 Materials and Methods 2.1

Electrode

The bipolar electrodes were constructed using a standard cooper printed circuit board based on the procedure describe by Ou [9]. The electrode was as pattern an external partial ring with 8 mm of inner diameter and 10 mm of outside diameter, and a center diameter of 1 mm (Fig. 1a). The electrode surface was covered using a mixture of epoxy Di-Glycidyl Ether Bisphenol A (DGEBA) and 2.0% weight of graphene sheets obtained by a modified Hummers method based on large oxidation period combined with a higher purification process.

Fig. 1. Experiment settings. (a) Bipolar electrode configuration showing geometrical structure, (b) Experiment setup showing the electrode inside the acrylic box (40 mm  40 mm  25 mm)

2.2

Impedance Measurements

Impedance measurements was performed by using HF2IS (Zürich Instruments - see Fig. 1b). It was set 1 Vpp, injected in the sample under study and collected 80 discrete frequencies between 100 Hz to 10 MHz. It was prepared 14 solutions of distilled water containing the following concentrations of glucose: 0, 36, 70, 100, 130, 158, 184, 208, 231, 252, 273, 531, 776, 1000 µM.

Biosensor Based on Carbon Nanocomposites

35

3 Results Figure 2 shows measured values of minimum reactance for each glucose concentration, the minimum reactance curve decreases with glucose concentration increase. Based on sensitivity behavior, the curve was divided in three zones. Zone 1 sets concentrations below 200 µM with a mean sensitivity of 348 Ω/µM. This zone shows a behavior adjusted to exponential with a Pearson coefficient of 0.9866. Zone 2 sets concentration between 200 to 500 µM with a sensitivity of 30.2 Ω/µM. The data was adjusted to a linear equation with a Pearson coefficient of 0.8382. Zone 3 sets above 500 µM has a mean sensitivity of 79.6 Ω/µM. The zone 3 was adjusted to a quadratic equation with a Pearson coefficient of 0.9901.

Fig. 2. Correlation plot between maximum reactance values and glucose concentration

Lower glucose concentration of 36 µM shows higher values in reactance and resistance (537 kΩ, −170 kΩ) compared with 1000 µM (210 kΩ, −54 kΩ). The ratio and minimum reactance peak decrease mutually with concentration increments. Contribution of Electrode Electrolyte Interface (EEI) in lower frequencies was more evident for concentrations higher than 531 µM.

4 Discussion The Nyquist plots brought coherent behavior for polarization theory in aqueous solution, lower concentrations of a solute in aqueous media always appear as a semicircle or semi-ellipsoid with higher reactance and resistance [10]. Semi-ellipsoid represents an asymmetric distribution, this indicates a dispersion of relaxation time and

36

J. A. Gomez-Sanchez et al.

therefore non-Debye relaxation type [11]. This electrochemical response of electrode is due to surface irregularities and caused by factors as: geometrical, chemical and nanostructure [12]. Electrode surface shows complex roughness and increase irregularities in nanometer scales, this complexity produces irregular interfaces and requires fractal models to understand [13]. The superficial nanoparticles increase the contact area between electrode and electrolyte, thus increasing the current density improving recombination by accelerating the electron conductivity. According to Nyquist plots, the transfer resistance and capacitance of double layer are higher than 50 kΩ, the system appears as kinetically slow due of unreactive plane of graphene [14]. A transition zone in between 200 to 500 µM was detected, this is due to electron transition processes in semiconductors such as graphene [4]. The minimum capacitive reactance values (Fig. 3) have a better sensitivity in glucose concentrations lower than 200 µM compared with others concentrations. This might be an advantage for measuring glucose concentration with sweat, saliva and urine [5]. Glucose monitors based on biosensor are more accurate than plasma laboratory analysis. Blood glucose can exhibit glucose peaks while interstitial glucose keeps stable resulting in fake peaks [15]. Making insulin corrections during these fake glucose peaks can result in a negative impact because glucose levels did not need to be reduced [16]. The DGEBA is considered as an endocrine disruptor; the use as a matrix in the electrode can influence the reactions with glucose. This reaction is possible due to the oxidative stress related to glucose metabolism [17].

5 Conclusion Based on our results, the sensor has higher sensitivity below 200 µM and above 500 µM. That is a great advantage to measure glucose concentration with nonconventional physiological fluids in hypoglycemic and hyperglycemic patients. Electrode Electrolyte Interface in low frequencies for lower concentrations is reduced due to nanoparticles that increased electrode surface area and unreactive plane of graphene. Acknowledgment. We thank CAPES for the financial support of this research, registered under the number 80072285940, FAPESC and UDESC for the institutional support.

References 1. Vlăsceanu, G.M., Amărandi, R.M., Ioniță, M., Tite, T., Iovu, H., Pilan, L., Burns, J.S.: Versatile graphene biosensors for enhancing human cell therapy. Biosens. Bioelectron. 9, 154–186 (2018) 2. Yan, F., Zhang, M., Li, J.: Solution-gated graphene transistors for chemical and biological sensors. Adv. Healthcare Mater. 3, 313–331 (2014) 3. Yina, H., Zhoub, Y., Menga, X., Shanga, K., Aia, S.: One-step “green” preparation of graphene nanosheets and carbon nanospheres mixture by electrolyzing graphite rob and its application for glucose biosensing. Biosens. Bioelectron. 30, 112–117 (2011)

Biosensor Based on Carbon Nanocomposites

37

4. Harper, A., Anderson, M.R.: Electrochemical glucose sensors—developments using electrostatic assembly and carbon nanotubes for biosensor construction. Sensors 10(9), 8248–8274 (2010) 5. Bruen, D., Delaney, C., Florea, L., Diamond, D.: Glucose sensing for diabetes monitoring: recent developments. Sensors 17, 1866 (2017) 6. Nitzan, O., Elias, M., Chazan, B., Saliva, W.: Urinary tract infections in patients with type 2 diabetes mellitus: review of prevalence, diagnosis, and management. Diabetes Metab. Syndr. Obes. 8, 129–136 (2015) 7. Lee, H., Song, C., Hong, Y.S., Kim, M.S., Cho, H.R., Kang, T., Shin, K., Choi, S.H., Hyeon, T., Kim, D.: Wearable/disposable sweat-based glucose monitoring device with multistage transdermal drug delivery module. Sci. Adv. 3(3), e1601314 (2017) 8. Diouf, A., Bouchikhi, B., El Bari, N.: A nonenzymatic electrochemical glucose sensor based on molecularly imprinted polymer and its application in measuring saliva glucose. Mater. Sci. Eng. C 98, 1196–1209 (2019) 9. Ou, J., Maldonado, A., Saephan, C., Farahmand, F., Caggiano, M.: A low-cost PCB fabrication process. In: 2014 IEEE 64th Electronic Components and Technology Conference (ECTC), Orlando, FL, pp. 2159–2162 (2014) 10. Urban, S., Unmüssig, T., Daubinger, P., Kieninger, J., Urban, G.: Stability of non-enzymatic glucose sensor based on platinum micro-/nanostructures. Proc. Eng. 120, 1145–1148 (2015) 11. Arya, A., Sharma, A.L.: Structural, electrical properties and dielectric relaxation in Na+ ion conducting solid polymer electrolyte. J. Phys. Condens. Matt. 30, 165402 (2018) 12. Atighilorestani, M., Brolo, A.G.: Comparing the electrochemical response of nanostructured electrode arrays. Anal. Chem. 89(11), 6129–6135 (2017) 13. Felice, C.J., Ruiz, G.A.: Differential equation of a fractal electrode–electrolyte interface. Chaos Solit. Fractals 84, 81–87 (2016) 14. Daniels, K.M., Shetu, S., Staser, J., Weidner, J., Williams, C., Sudarshan, T.S., Chandrashekhar, M.V.S.: Mechanism of electrochemical hydrogenation of epitaxial graphene. J. Electrochem. Soc. 162, E37 (2015) 15. Baek, Y.H., Jin, H.Y., Lee, K.A., Kang, S.M., Kim, W.J., Kim, M.G., Park, J.H., Chae, S. W., Baek, H.S., Park, T.S.: The correlation and accuracy of glucose levels between interstitial fluid and venous plasma by continuous glucose monitoring system. Korean Diabetes J. 34(6), 350–358 (2010) 16. Cengiz, E., Tamborlane, W.: A tale of two compartments: interstitial versus blood glucose monitoring. Diabetes Technol. Therap. 11, S11–S16 (2009) 17. Moghaddam, H.S., Samarghandian, S., Farkhondeh, T.: Effect of bisphenol A on blood glucose, lipid profile and oxidative stress indices in adult male mice. Toxicol. Mech. Methods 25(7), 507–513 (2015)

Bioimpedance Measurements on Human Neural Stem Cells as a Benchmark for the Development of Smart Mobile Biomedical Applications André B. Cunha1(&), Christin Schuelke1(&), Arto Heiskanen2, Afia Asif2, Yasmin M. Hassan3, Stephan S. Keller3, Håvard Kalvøy4, Alberto Martínez-Serrano5, Jenny Emnéus2, and Ørjan G. Martinsen1,4 1

Department of Physics, University of Oslo, Sem Sælands vei 24, 0371 Oslo, Norway {andre.cunha,christin.schuelke}@fys.uio.no 2 Technical University of Denmark, DTU Bioengineering, 2800 Kongens Lyngby, Denmark 3 Technical University of Denmark, DTU Nanolab, 2800 Kongens Lyngby, Denmark 4 Department of Clinical and Biomedical Engineering, Oslo University Hospital, Sognsvannsveien 20, 0372 Oslo, Norway 5 Department of Molecular Neurobiology, Center of Molecular Biology ‘Severo Ochoa’, Universidad Autónoma de Madrid, Calle Nicolás Cabrera 1, 28049 Madrid, Spain

Abstract. Over the past 30 years, stem cell technologies matured from an attractive option to investigate neurodegenerative diseases to a possible paradigm shift in their treatment through the development of cell-based regenerative medicine (CRM). Implantable cell replacement therapies promise to completely restore function of neural structures possibly changing how we currently perceive the onset of these conditions. One of the major clinical hurdles facing the routine implementation of stem cell therapy is the limited and inconsistent benefit observed thus far. While unclear, numerous pre-clinical and a handful of clinical cell fate imaging studies point to poor cell retention and survival. Coupling the need to better understand these mechanisms while providing scalable approaches to monitor these treatments in both pre-clinical and clinical scenarios, we show a proof of concept bioimpedance electronic platform for the Agile development of smart and mobile biomedical applications like neural implants or highly portable monitoring devices. Keywords: Electrical impedance spectroscopy (EIS)
Bioimpedance
Neural stem cells (NSCs)
Proliferation
Embedded system
Mobile technologies

A. B. Cunha and C. Schuelke—Both authors contributed equally to this work. © Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 38–47, 2020. https://doi.org/10.1007/978-981-13-3498-6_6

Bioimpedance Measurements on Human Neural Stem Cells

39

1 Introduction Around 1905, the electric galvanometers were sensitive and rapid enough to noninvasively pick up the very small potential difference generated by heart activity (ECG), and, by 1910, Höber used bioimpedance methods to prove the existence of cell membranes and calculated how extremely thin they are (less than 0.01 µm) [1]. Since then, many clinical applications based upon bioimpedance have been developed, including measuring cardiac minute volume non-invasively, measuring lung respiration activity, and taking electronic biopsies for diagnosis of skin cancer among others [2]. More than 100 years later, as we enter the age of cell-based regenerative medicine (CRM), bioimpedance is more relevant than ever. Reasonable progress has been made within the field of stem cell therapy, but there is still a limited and inconsistent benefit provided by numerous pre-clinical and a handful of clinical CRM approaches. While the challenges like poor cell retention and survival or teratoma [3, 4] of pluripotent stem cells are well known effects in vivo, the inability to determine cell fate and survival in humans has been a significant obstacle to understanding the mechanisms of the variable efficacy. While the incorporation of cell fate imaging in clinical trials may help address these significant hurdles, live bioimpedance surveillance may be the perfect candidate to monitor those treatments. Moreover, multifaceted monitoring and control strategies will be vital in addressing these issues to enable the successful rise of cell-based regenerative medicine. Numerous studies based on impedance measurements of live biological cells enabled the technique to become widely accepted as a label free, non-invasive and quantitative analytical method to assess cell status. To show some examples, bioimpedance can be used to monitor proliferation [5], apoptosis [6], migration [7], degeneration [8], morphological changes [9] and also (neuronal) differentiation [10]. In this work, we show a proof of concept bioimpedance electronic platform for the Agile development of smart and mobile biomedical applications like neural implants or highly portable monitoring devices. Most current bioimpedance setups rely on what we will define as a traditional approach. A typical traditional approach in portable medical electronic system comprises components like analog front-ends for data acquisition, amplifiers and filters for signal conditioning, analog to digital (ADC) converters for signal and sensor data acquisition, buttons to accept user feedback, an MCU to execute algorithms, and a variety of interfaces such as an LCD display, USB port and so on. Traditional design methodologies bring together all of the needed components onto a printed circuit board (PCB) [11]. Modern system-on-chip (SoC) architectures provide a new way of designing portable medical electronic devices bringing numerous advantages. Portable medical electronics equipment of all types, like glucose meters, pulse oximeters, portable ECG devices, etc. – are already implemented using SoCs [11]. By integrating many of the peripheral components required by portable medical electronics applications, fewer components are needed resulting in simpler PCBs which take less time to prototype

40

A. B. Cunha et al.

resulting in shorter iterations. In the case of PSoC (programmable SoC) superior reconfigurability greatly reduces the need for new prototypes for each iteration. By speeding up iterations [12] and making prototypes more flexible, PSoC development platforms enable the usage of Agile development methodologies. To show a proof of concept of our PSoC-based bioimpedance analyzer, electrical impedance spectroscopy was performed by using both our proposed instrument in comparison to a commercial impedance analyzer to assess proliferation of the neural stem cell line hVM1 (human ventral mesencephalic neural stem cell line 1).

2 Materials and Methods 2.1

Development Methodology and Contextual Framework

We show a proof of concept for the bioimpedance measurement instrumentation to be used on the Training4CRM (European Training Network for Cell-based Regenerative Medicine [13]) optogenetic neural implant for the treatment of neurodegenerative diseases. Apart from the control logic core, the implant will roughly consist of optogenetically modified neuronal cells, a sensing fiber electrode, and an analog front-end, both still under parallel development. Interfacing biomedical and engineering technologies under parallel development is one of the major challenges. To handle this, an open Agile iterative development methodology was adopted. To overcome the restraints of traditional electronics hardware development, the Cypress PSoC (programmable system-on-chip) 32-bit Arm® Cortex®-M3 PSoC® 5LP platform was chosen. These chips include a CPU core and mixed-signal arrays of configurable integrated analog and digital peripherals. This hybrid architecture enables high flexibility for quick developing and prototyping as a tradeoff for state-of-the-art performance allowing the translation of many hardware design decisions to software development choices. The current implementation is built upon the CY8CKIT-050 PSoC® 5LP Development Kit and external passive components using concepts based on application notes suggested in Cypress online resources. 2.2

Analog and Digital Design

As it can be seen in the schematic in Fig. 1, the current implementation is based on a four-electrode configuration. However, for the sake of this work, this system was adapted to a two-electrode configuration by connecting the electrodes. The excitation source is an 8-bit current output DAC (digital to analog converter) which can source or sink current in three ranges (2040 lA, 255 lA, and 31.875 lA). For this application, the source was set at 31.875 lA.

Bioimpedance Measurements on Human Neural Stem Cells

41

Fig. 1. Analog design schematic of PSoC-based impedance analyzer

A PWM (pulse width modulator) generates the clock to trigger the DMA (direct memory access) transactions which transfer a sine wave LUT (lookup table) to the current output DAC every quarter wave and changing the sign though source and sink control every half wave. The frequency excitation signal can be changed through the period of the PWM divider. This approach limits the number of available frequencies and the DMA transfer speed creates a bottleneck for the maximum available excitation frequency even using a low-resolution waveform. Tuning these variables through iteration, we generate 10 frequencies from 100 Hz to 100 kHz in multiples of 1, 2 and 5. Each quarter of wave table has 13 elements with bigger tables resulting in a significant slower maximum frequency due to the DMA bottleneck. Future iterations can use alternative approaches exploiting the FPGA (field-programmable gate array) abilities of the platform to overcome these limitations. However, for the purpose of establishing a proof of concept device, they are sufficient. Finally, an internal operational amplifier is used as shown in the schematic in Fig. 1 as a low pass filtered current to voltage converter. The RD ADC (analog to digital converter) modulator input is used to dynamically control the polarity of the signal in the modulator inverting it when the input is high. This way, it can be used to modulate the input signal with an independent clock to act as a signal mixer. Two state machines control both the ADC and also generate in-phase and quadrature clock signals to be used in the signal mixer. 2.3

Embedded Application

The firmware was kept as simple as possible. In its current form, the development kit allows the device to work standalone requiring external power only. Through the integrated LCD, the user is able to read the measurements for both phase and amplitude while a button cycles through the available frequencies by setting the divider of the PWM. The application itself uses a loop based architecture. In a simplified description, the routine starts by initializing the necessary hardware namely the operational amplifiers and calibrating them, and then it initializes the excitation signal generator and DMA, the ADC and finally the LCD. The measurement routine follows by getting in-phase and in-quadrature measurements for the transimpedance amplifier but also for

42

A. B. Cunha et al.

the excitation signal selectable through two multiplexers. Phase and amplitude for impedance are calculated through synchronous detection and they are then printed in the LCD. 2.4

Cell Culture of hVM1

hVM1 Cell Line: hVM1 is a human ventral mesencephalic neural stem cell line. Cell isolation and immortalization were described previously [14]. Briefly, human neural stem cells (hNSCs) were isolated from a 10-week-old aborted fetus (Lund University Hospital). Tissue procurement was in accordance with the Declaration of Helsinki and in agreement with the ethical guidelines of the European Network of Transplantation. The cells were immortalized by stable transfection with a retroviral vector coding for v-myc (LTR-vmyc-SV40p-Neo-LTR) [15]. It has been shown that neuronal differentiation results predominantly in dopaminergic neurons that exhibit electrophysiological activity [16]. Stable transfection with the vector LTRBcl-XL-IRES-rhGFP-LTR resulted in expression of the anti-apoptotic protein Bcl-XL (basal cell lymphoma – extra-large) which increased the dopaminergic properties [17, 18]. Cells were routinely cultured in cell culture flasks pre-treated with Geltrex™ (ThermoFisher Scientific, A1413301) and grown at 37 °C and 5% CO2. hVM1 Growth Medium: Dulbecco’s modified Eagle medium/F-12 medium with Glutamax (ThermoFisher Scientific) supplemented with 0.5% Albumax (ThermoFisher Scientific), 5 mM HEPES (ThermoFisher Scientific), 0.6% glucose (Sigma), N2 supplement (ThermoFisher Scientific), non-essential amino acids (Ala, Asn, Asp, Glu, Pro, 40 mM each; MerckMillipore), penicillin/streptomycin, epidermal growth factor and basic fibroblast growth factor (20 ng/ml each; R&D Systems). 2.5

Carbon Electrode Chips

The pyrolytic carbon electrode chips were provided by Technical University of Denmark [19]. They are composed of a circular pyrolytic carbon working electrode (WE) with an area of 12.5 mm2, surrounded by a platinum counter electrode (CE) with an area of 25.2 mm2. The reference electrode (RE) has an area of 0.8 mm2 and is also made from Pt, as well as the electrical contacts and leads. The chip (10 mm  30 mm) is finally covered by an insulating SU-8 passivation layer, except for the sensing area and the contact pads (Fig. 2). The electrode chips were initially treated with oxygen plasma and then assembled in a chip holder with magnetic clamping. The wells were sterilized using 0.5 M NaOH solution for 15 min, followed by washing with cell-culture-tested water. After coating of the electrode surface with Geltrex for at least 2 h, hVM1 cells were seeded at a density of 0.4 Mio cells/cm2 and cultivated at 37 °C and 5% CO2.

Bioimpedance Measurements on Human Neural Stem Cells

43

Fig. 2. Schematic drawing (left) of 2D carbon electrodes: layer of pyrolytic carbon (A), platinum electrodes, contact leads and pads (B), SU-8 passivation layer (C), modified after [19]. CE = counter electrode, WE = working electrode, RE = reference electrode. Photo of electrodes assembled in chip holder with magnetic clamping (right).

2.6

Electrical Impedance Spectroscopy

At this point, one of the main goals is to verify the viability of the measurement principle of the PSoC system while ensuring the sampled cells survive the excitation signal. Therefore, the system under test is solely used to measure on the first well of the triple chip system, while our reference system, a Zurich Instruments MFIA Impedance Analyzer is used on the three wells to both assess cell growth and also provide a reference to the measurements done with our proposed instrumentation. The MFIA connects to the chip holder using a couple of 1 m long RG-6 coaxial cables while our proposed instrument uses 8 cm 0.5 mm wide copper wires. For the excitation signal, the MFIA uses a sinusoidal voltage source of 45.78 µV with working currents ranging up to 200 nA in the chosen frequency range while our instrument uses a 32 mV source from the 32 µA AC source, resulting in estimated currents around 300 µA. In broad terms, both instruments use the lock-in amplifier technique. This technique has been widely used in impedance measurement applications being the gold standard in noise suppression and accuracy. Impedance spectroscopy was performed every second day to follow the proliferation of hVM1 cells for the total duration of 9 days. A two-electrode configuration was used by connecting only working and counter electrode. A medium change was performed before each measurement. Spectra from 1 kHz to 1 MHz (100 points, log) were recorded with the MFIA, while with the PSoC prototype, a spectrum from 100 Hz to 100 kHz (10 points) was analyzed. After the end of the experiment, cells were removed from the electrode surface with trypsin and an impedance spectrum was recorded of the electrode system without cells, but with the same amount of medium providing a reference. Due to electrode degradation, it was not possible to do a no cells reference measurement using the PSoC instrument.

44

A. B. Cunha et al.

3 Results and Discussion 3.1

Characterization of Cell Proliferation by Electrical Impedance Spectroscopy (EIS)

Bioimpedance measurements to follow the proliferation of hVM1 cells have been performed to assess the performance of the PSoC-based impedance analyzer compared to a commercial instrument, the Zurich Instruments MFIA Impedance Analyzer. The resulting spectra of impedance and phase angle as a function of frequency are shown in Fig. 3. For the PSoC-based instrument, impedance values could only be measured up to 100 kHz due to limitations of the platform. Due to electrode deterioration, no measurement without cells was possible. From both the commercial instrument and the developed impedance analyzer, it is visible that the impedance increases with the cultivation period. An increasing number of cells leads to an increased impedance [20]. No significant change of the phase angle h was detectable.

100 1.0 10 3

1.0 10 4

1.0 10 5

B

-60

Theta [°]

200

|Z| [

300

]

Zurich Instruments MFIA A

-40

-20

0 1.0 10 3

1.0 10 6

d9 d7 d5 d3 d1 no cells

1.0 10 4

1.0 10 5

1.0 10 6

Frequency [Hz]

Frequency [Hz]

100 1.0 10 3

1.0 10 4

1.0 10 5

Frequency [Hz]

1.0 10 6

D

-60

Theta [°]

200

|Z| [

300

]

PSoC-based instrument C

-40

d9 d7 d5 d3 d1

-20

0 1.0 10 3

1.0 10 4

1.0 10 5

1.0 10 6

Frequency [Hz]

Fig. 3. EIS with Zurich Instruments MFIA (A, B) and PSoC-based impedance analyzer (C, D). Impedance |Z| (A, C) and phase angle h (B, D) as a function of frequency for different days of growth of hVM1 cells (d1–d9) in comparison to the electrode system without cells (only MFIA). No cells could not be shown for PSoC due to electrode deterioration. MFIA n = 3, dotted line represents SEM. PSoC n = 1.

##

relative impedance [%]

130

###

* ** 120

**

110

relative impedance [%]

Bioimpedance Measurements on Human Neural Stem Cells 120 115 110 105 100 100

d1

d3

d5

d7

d9

45

d3

d5

d7

d9

Zurich Instruments PSoC-based Instrument

Fig. 4. Relative impedance [%] for growth of hVM1 cells normalized to electrode chip without cells (left, n = 3) or d1 due to electrode deterioration (right, n = 1) for Zurich Instruments vs. PSoC-based instrument. * p < 0.05, ** p < 0.01, *** p < 0.001. # indicates significant difference to system without cells.

The normalization of impedance values for the different days of growth to the impedance of the chip without cells provides a quantification of the change in impedance that is caused by the cells only, eliminating the influence of electrode design, cables etc. This relative impedance was calculated using Eq. 1 and the maximum over the frequency range is shown in Fig. 4. relative impedance ¼

jZjwith cells
100% jZjwithout cells

ð1Þ

The calculation of relative impedance confirms the increasing impedance values due to cell growth. From day 7, the growth of the cells resulted in a significantly higher impedance compared to the system without cells. This trend is also visible with the PSoC-based instrument. However, the measured increase in impedance is smaller for the PSoC-based instrument than for the MFIA impedance analyzer. The single well 10 point measurements done with the PSoC is not enough to assess the reason for this behavior requiring further work to understand the underlying causes.

4 Conclusions The instrument proposed in this article provides a miniaturized, highly flexible and affordable solution in a single-chip format allowing for impedance measurements of adherent cells. A single chip fully integrated solution provides superior immunity to noise, allowing fast development of miniaturized prototypes. A PSoC integrated circuit is composed of a core, configurable analog and digital blocks, and programmable routing and interconnect. The configurable blocks in a PSoC are the biggest difference from other microcontrollers. In this work, we exploited this ability to use a RD ADC as a mixer for achieving synchronous detection. The results suggest that the applied current does not harm the cells and that the system allows for measurement of cell

46

A. B. Cunha et al.

growth comparable to a commercial impedance analyzer. However, future work aims at improving both excitation and sensitivity systems, as well as testing our proposed instrument in bioimpedance measurements of other cellular phenomena. Acknowledgment. This project has been funded by the European Union Horizon 2020 Programme (H2020-MSCA-ITN-2016) under the Marie Skłodowska-Curie Innovative Training Network and Grant Agreement No.722779. Conflict of Interest. The authors declare that they have no conflict of interest.

References 1. McAdams, E.T., Jossinet, J.: Tissue impedance: a historical overview. Physiol. Meas. 16 (3A), A1. https://doi.org/10.1088/0967-3334/16/3a/001 2. Grimnes, S., Martinsen, Ø.G.: Bioimpedance and Bioelectricity Basics, 3rd edn. Academic Press is an Imprint of Elsevier, London (2015) 3. Ronaghi, M., Erceg, S., Moreno-Manzano, V., et al.: Challenges of stem cell therapy for spinal cord injury: human embryonic stem cells, endogenous neural stem cells, or induced pluripotent stem cells? Stem Cells 28(1), 93–99 (2010). https://doi.org/10.1002/stem.253 4. Nguyen, P.K., Neofytou, E., Rhee, J.-W., et al.: Potential strategies to address the major clinical barriers facing stem cell regenerative therapy for cardiovascular disease: a review. JAMA Cardiol. 1(8), 953–962 (2016). https://doi.org/10.1001/jamacardio.2016.2750 5. Xiao, C., Luong, J.H.T.: On-line monitoring of cell growth and cytotoxicity using electric cell-substrate impedance sensing (ECIS). Biotechnol. Prog. 19(3), 1000–1005 (2003). https://doi.org/10.1021/bp025733x 6. Krinke, D., Jahnke, H.-G., Mack, T.G.A., et al.: A novel organotypic tauopathy model on a new microcavity chip for bioelectronic label-free and real time monitoring. Biosens. Bioelectron. 26(1), 162–168 (2010). https://doi.org/10.1016/j.bios.2010.06.002 7. Hug, T.S.: Biophysical methods for monitoring cell-substrate interactions in drug discovery. Assay Drug Dev. Technol. 1(3), 479–488 (2003). https://doi.org/10.1089/ 154065803322163795 8. Jahnke, H.-G., Braesigk, A., Mack, T.G.A., et al.: Impedance spectroscopy based measurement system for quantitative and label-free real-time monitoring of tauopathy in hippocampal slice cultures. Biosens. Bioelectron. 32(1), 250–258 (2012). https://doi.org/10.1016/j.bios. 2011.12.026 9. Haas, S., Jahnke, H.-G., Glass, M., et al.: Real-time monitoring of relaxation and contractility of smooth muscle cells on a novel biohybrid chip. Lab Chip 10(21), 2965–2971 (2010). https://doi.org/10.1039/c0lc00008f 10. Seidel, D., Obendorf, J., Englich, B., et al.: Impedimetric real-time monitoring of neural pluripotent stem cell differentiation process on microelectrode arrays. Biosens. Bioelectron. 86, 277–286 (2016). https://doi.org/10.1016/j.bios.2016.06.056 11. Kumar, S.: Reducing complexity and cost for portable medical electronics through system on chip architectures (2010). https://www.cypress.com/file/102926/download. Accessed 09 Jan 2020 12. Saunders, M.: Software development models for PSoC 6 (2017). https://www.cypress.com/ blog/technical/software-development-models-psoc-6. Accessed 09 Jan 2020

Bioimpedance Measurements on Human Neural Stem Cells

47

13. CORDIS: European Training Network for Cell-based Regenerative Medicine | Projects | H2020 | CORDIS | European Commission (2019). https://cordis.europa.eu/project/rcn/ 205439/factsheet/en. Accessed 28 Feb 2019 14. Villa, A., Liste, I., Courtois, E.T., et al.: Generation and properties of a new human ventral mesencephalic neural stem cell line. Exp. Cell Res. 315(11), 1860–1874 (2009). https://doi. org/10.1016/j.yexcr.2009.03.011 15. Villa, A., Snyder, E.Y., Vescovi, A., et al.: Establishment and properties of a growth factordependent, perpetual neural stem cell line from the human CNS. Exp. Neurol. 161(1), 67–84 (2000). https://doi.org/10.1006/exnr.1999.7237 16. Tønnesen, J., Seiz, E.G., Ramos, M., et al.: Functional properties of the human ventral mesencephalic neural stem cell line hVM1. Exp. Neurol. 223(2), 653–656 (2010). https:// doi.org/10.1016/j.expneurol.2010.01.013 17. Krabbe, C., Courtois, E., Jensen, P., et al.: Enhanced dopaminergic differentiation of human neural stem cells by synergistic effect of Bcl-xL and reduced oxygen tension. J. Neurochem. 110(6), 1908–1920 (2009). https://doi.org/10.1111/j.1471-4159.2009.06281.x 18. Courtois, E.T., Castillo, C.G., Seiz, E.G., et al.: In vitro and in vivo enhanced generation of human A9 dopamine neurons from neural stem cells by Bcl-XL. J. Biol. Chem. 285(13), 9881–9897 (2010). https://doi.org/10.1074/jbc.M109.054312 19. Hassan, Y.M., Caviglia, C., Hemanth, S., et al.: High temperature SU-8 pyrolysis for fabrication of carbon electrodes. J. Anal. Appl. Pyrolysis 125, 91–99 (2017). https://doi.org/ 10.1016/j.jaap.2017.04.015 20. Witzel, F., Fritsche-Guenther, R., Lehmann, N., et al.: Analysis of impedance-based cellular growth assays. Bioinformatics 31(16), 2705–2712 (2015). https://doi.org/10.1093/ bioinformatics/btv216

Bioimpedance Theory and Modelling

Numerical Simulation of Various Electrode Configurations in Impedance Cardiography to Identify Aortic Dissection Alice Reinbacher-Köstinger1(&), Vahid Badeli1, Gian Marco Melito2, Christian Magele1, and Oszkar Bíró1 1

2

Institute of Fundamentals and Theory in Electrical Engineering, Graz University of Technology, 8010 Graz, Austria [email protected] Institute of Mechanics, Graz University of Technology, 8010 Graz, Austria

Abstract. Impedance cardiography (ICG) is a non-invasive method to evaluate several cardiodynamic parameters. Pathologic changes in the aorta, like an aortic dissection, will alter the aortic shape as well as the blood flow and, consequently, the impedance cardiogram. This fact distorts the evaluated cardiodynamic parameters on the one hand, and offers the possibility to identify aortic pathologies on the other hand. In order to find an appropriate measurement configuration, in particular for the identification of aortic dissections, a 3D simulation model has been used. Various electrode positions have been investigated to reach a high sensitivity with respect to the discrepancy between the healthy and the dissected case. Keywords: Impedance cardiography simulation


Aortic dissection
Numerical

1 Introduction Aortic dissection (AD) is a serious condition in which the inner layer of the aorta tears. Consequently, blood starts to flow within the media layer and develops a so called false lumen. This condition may become acute with a high mortality rate within a few hours or chronic, which means that the onset of the dissection dates back more than 14 days and patients can be often treated with medical therapy. In both cases an easy to use and still reliable non-invasive method to identify the presence of an AD would be beneficial. Impedance Cardiography is such a non-invasive method, which is already in clinical use. By injecting a low-amplitude alternating current into the thorax, impedance changes can be identified during the cardiac pulse wave by measuring the voltage drop. Since the conductivity of the blood filled aorta is much higher than that of the surround-ing tissue types, changes of the measured impedance are strongly related to changes in the aorta. Studies have shown that there are two main reasons for temporal impedance changes: one results from the volume change of the aorta due to higher pressure in the systolic phase, i.e. when the heart ventricles contract and thus © Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 51–54, 2020. https://doi.org/10.1007/978-981-13-3498-6_7

52

A. Reinbacher-Köstinger et al.

force blood into the aorta. The second one has its origin in blood flow velocity, because the red blood cells get oriented and deformed at higher flow rates, which increases the conductivity of blood. Since both the blood volume and the flow will change in case of an aortic dissection, an altered impedance cardiogram can be expected. In [1] the differences of the impedance curve between the healthy and the dissected case have already been investigated using a 3D simulation model. The results seem to be promising, nevertheless, for the identification of aortic dissections the sensitivity of the method has to be improved.

2 Methods 2.1

Simulation Model

A 3D numerical simulation model has been set up in COMSOL Multiphysics software [2] to investigate the changes in the electric potential and, furthermore, the impedance changes on the thorax surface. The geometry, formulation and the material properties are the same as described in the companion paper in this proceedings book [3]. Additional potential drops are calculated on the front and the back of the thorax at the same z-level of Vtop and Vbottom (Fig. 1).

Fig. 1. 3D simulation model setup.

Based on the formulations defined in [4], the blood conductivity rbl has been calculated analytically as a function of the spatial average velocity as described in [3] for several time instants in the simulation model. In case of an aortic dissection, the flow is highly disturbed locally and changes to a turbulent flow with strong recirculation [5]. Assuming this, the red blood cells will not be aligned and deformed like in the case of laminar flow. Thus, the conductivity of blood in the aorta will not change

Numerical Simulation of Various Electrode Configurations in ICG to Identify AD

53

during a pulse cycle like in the non-dissected (or healthy) case and can be assumed to be constant over time for both the true and the false lumen.

3 Results For all simulations one pair of injecting electrodes and three pairs of sensor electrodes have been used in the model. While the positions of the sensor electrodes are fixed (front, back and side), various injection electrode positions have been investigated. To verify the sensitivity with respect to the false lumen position, also two different false lumen locations in the xy plane has been assumed (see the dashed circle in Fig. 2). In Fig. 2 the impedance cardiogram djZ ðtÞj=dt is shown for the healthy and the aortic dissection case, for the three sensor electrode pairs separately. It can be observed that the maximum value of djZ ðtÞj=dt, which is used to evaluate the stroke volume in ICG, is lower in the aortic dissection case, in particular at the front and the back sensor electrodes. The side sensor electrodes provide a higher absolute value, while the difference between the healthy and the dissected case is lower. In Table 1 the decrease of djZ ðtÞj=dtjmax has been quantified by evaluating the relative change based on the healthy case. Obviously, the sensitivity is low when the injection is at the same position as the sensors, since only a small region is covered. Besides the clear discrepancy of the djZ ðtÞj=dtjmax value, also the minimal value of the impedance cardiogram and its time shift is altered unambiguously in the dissected case.

sensor electrodes back shifted injection tl

fl

side

y x

front

Fig. 2. Impedance cardiograms and sketch of measurement configuration assuming shifted injection. Only true lumen (tl) and false lumen (fl) are shown, but the surrounding tissue is also considered in the simulations. The dashed circle depicts the shifted false lumen.

54

A. Reinbacher-Köstinger et al.

Table 1. Relative difference of djZ ðtÞj=dtjmax between the healthy and the dissected case for each sensor electrode pair separately. 6 variants of injection electrode position (inj.) and location of the false lumen (fl) have been investigated. Inj. Inj. Inj. Inj. Inj. Inj.

Side side, fl side 5.90% side, fl shifted 13.65% back, fl side 46.27% back, fl shifted 68.02% shifted, fl side 45.42% shifted, fl shifted 59.63%

Back 50.76% 54.45% 32.79% 48.13% 54.74% 69.82%

Front 49.08% 56.77% 54.06% 77.55% 57.38% 71.10%

Mean 35.25% 41.63% 44.37% 64.68% 52.45% 66.85%

4 Discussion and Conclusion Different electrode positions have been investigated with a 3D numerical simulation model to reach the highest sensitivity in order to identify the presence of an aortic dissection by ICG. Basically, three pairs of sensor electrodes have been assumed to consider additional spatial information, which will be needed because of the unknown position of the false lumen. It can be observed that the configuration with the shifted injection electrodes features the highest mean sensitivity, independent from the investigated false lumen position. Additional information regarding the identification of an aortic dissection can be found in the shifted left ventricular ejection time (tLVET) and in the minimal value of the impedance cardiogram. Acknowledgment. This work is part of the LEAD project “Mechanics Modeling and Simulation of Aortic Dissection”, funded by Graz University of Technology.

References 1. Reinbacher-Köstinger, A., et al.: Numerical simulation of conductivity changes in the human thorax. IEEE Trans. Magn. 55, 1–4 (2019) 2. COMSOL multiphysics v. 5.3. COMSOL AB, Stockholm, Sweden. www.comsol.com 3. Badeli, V., et al.: Numerical simulation of impedance cardiogram changes in case of chronic aortic dissection, companion paper. In: Proceedings of ICEBI (2019) 4. Hoetink, A.E., et al.: On the flow dependency of the electrical conductivity of blood. IEEE Trans. Biomed. Eng. 51(7), 1251–1261 (2004) 5. Cheng, Z., et al.: Analysis of flow patterns in a patient-specific aortic dissection model. J. Biomech. Eng. 132(5), 051007 (2010). https://doi.org/10.1115/1.4000964

Numerical Simulation of Impedance Cardiogram Changes in Case of Chronic Aortic Dissection Vahid Badeli(&), Alice Reinbacher-Köstinger, Oszkar Biro, and Christian Magele Institute of Fundamentals and Theory in Electrical Engineering, Graz University of Technology, 8010 Graz, Austria [email protected] Abstract. Aortic dissection is an extremely dangerous aortic disease which alters the aortic shape as well as the blood flow in the region concerned. A numerical simulation model based on the Thoracic Electrical Bioimpedance technique for investigating effects caused by the aortic dissections is proposed. The effect of these changes on time-dependent hemodynamic parameters are shown. Keywords: Thoracic electrical bioimpedance Non-invasive measurement method


Aortic dissection


1 Introduction Aortic dissection (AD) is initiated by an intimal tear which further penetrates through the medial layer, resulting in separation of the aortic wall layers and subsequent formation of a true lumen and a false lumen, Fig. 1. The presence of a false lumen alters the aortic hemodynamics and also changes the tissue distribution in the thorax. These changes can be basically identified and quantified by Impedance Cardiography (ICG). An investigation on a certain aortic dissection type B by applying the ICG method is done in this work.

2 Method Changes of the measured impedance are strongly related to changes in the aorta, such as increased volume in the systolic phase and blood flow induced conductivity changes. In case of an aortic dissection, the blood volume distribution as well as the blood flow profile changes compared to the healthy state. While the changes of the blood volume distribution can be easily modeled by adapting the geometry, a closer look at the flow-induced conductivity alteration of blood is necessary. The change of the electrical conductivity of blood due to flow has been discussed in a previous work [1]. Based on the method described in [1], the blood conductivity rbl has been calculated analytically as a function of the spatial average velocity as well as a function of the average radius of the aorta. Figure 2 clearly shows that the blood flow velocity highly influences the conductivity of blood, while the changing of the aortic radius within the considered range has a very low impact on it. © Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 55–59, 2020. https://doi.org/10.1007/978-981-13-3498-6_9

56

V. Badeli et al.

ascending aorta descending aorta

False lumen

Fig. 1. Descending aortic dissection

3 Simulation Model A 3D numerical simulation model with simple geometry, has been set up in COMSOL Multiphysics software [2], for the underlying time-harmonic current flow problem, Fig. 3. The source electrodes are placed on the side of the thorax and inject an AC current with a magnitude of 5 mA and a frequency of 100 kHz. The electric potential drop is evaluated between the measuring electrodes by solving the partial differential equation for the electric potential V, (1), which leads to the thoracic impedance, (2). Dissimilar to conventional ICG methods, propitious electrodes positions are chosen to eventuate the most noticeable difference between measurement results of healthy and dissected cases [3]. Material properties are assigned to the tissues based on [4]. For other surrounding materials which are not considered in the model, a mean conductivity and permittivity is assigned to the thorax domain to provide a realistic value for the thoracic base impedance jZ0 j of about 20 X as reported in [5]. As dynamical changes of lung perfusion and heart contraction nearly compensate themselves [5], only the conductivity changes of the blood and the volumetric changes of the aorta are considered. Measurement values of the blood flow velocity and the cross section area of a healthy aorta [6] are used as time-dependent input variables in the simulation model, Fig. 2. For a similar type of aortic dissection [7], as considered in this work, it has been shown that 60% of flow passes the true lumen and 40% passes the false lumen. The average blood flow velocity for each lumen can be calculated based on this assumption and Consequently, the blood conductivity changes for the false lumen and the true

Fig. 2. (Left) Blood conductivity dependency on the blood flow velocity and the radius of the aorta. (Right) Blood conductivity and aortic radius changes over a heart cycle for a healthy aorta.

Numerical Simulation of ICG Changes in Case of Chronic AD

57

Fig. 3. Simulation model setup.

lumen. However, it has also been shown that in case of an aortic dissection the flow is highly disturbed locally and changes to a turbulent flow with strong recirculation [8]. Assuming this, the red blood cells may not be aligned and deformed like in the laminar flow [1]. Thus, the conductivity of blood in the aorta will not change like in the healthy case. Nevertheless, since no experimental or simulation data exist regarding conductivity changes in turbulent flow, three different study cases have been investigated: (i) healthy case with time-varying blood conductivity (rbl ðtÞ), (ii) dissected case 1, considering a false lumen and laminar flow, which leads to a lower blood conductivity (rbl ðtÞ) in the true lumen, compared to the healthy case, and (iii) dissected case 2, considering turbulent flow and therefore, a constant blood conductivity (rbl ).

4 Results All the simulations have been performed at 21 discrete time instants for the 3 different study cases described in the previous section. In dissected cases, the radius of the false lumen is 15 mm, while the radius of the true lumen is dynamically changing (Fig. 2). A comparison of impedance changes during a pulse cycle for the 3 study cases is shown in Fig. 4. It has to be mentioned that the base impedance jZ0 j is different for each person. It is also different for the healthy and the dissected cases because of presence of a false lumen in the latter [1]. Hence discrepancies between the three cases are more visible by showing the impedance changes (jZ ðtÞj  jZ0 j) only. As expected, the impedance changes in all cases are different since different blood conductivity profile have been assumed in the aorta. The impedance changes in the dissected case 2 happens because of the blood volume variation while in the healthy case and the dissected case 1, the blood conductivity is changing by velocity as well.

58

V. Badeli et al.

Fig. 4. (Left) Impedance changes over a cycle. (Right) ICG signals for 3 different study cases.

In Fig. 4, ICG signals i.e.,  dZ dt , for the 3 study cases are depicted. It is shown in [5] that the blood flow-induced conductivity changes together with the volumetric changes of the aorta play a major role in forming the ICG signal. The ICG signals for the 3 study cases in this work, show a good agreement to that and lead to different timedependent hemodynamic parameters like the stroke volume (SV) and the cardiac output (CO). Thus a wrong diagnosis will be made. As it is expected, when the average blood flow velocity tends to zero, changes in the two dissected cases are the same. Furthermore, the difference between the healthy and the dissected cases is due to different blood conductivity profiles during the pulse cycle.

5 Conclusion A 3D simulation model has been set up to investigate the effect of an aortic dissection on the impedance changes on the thorax surface. A healthy and two different dissected cases have been investigated. Different impedance changes during a pulse cycle as well as different ICG signals will result in different time-dependent hemodynamic parameters and show the feasibility of diagnosing aortic dissection by this method. Acknowledgment. This work is part of the LEAD project Mechanics Modeling and Simulation of Aortic Dissection, funded by Graz University of Technology.

References 1. Reinbacher-Kostinger, A., et al.: Numerical simulation of conductivity changes in the human thorax caused by aortic dissection. IEEE Trans. Magn. 55(6), 1–4 (2019) 2. COMSOL multiphysics v. 5.3. COMSOL AB, Stockholm, Sweden. www.comsol.com 3. Reinbacher-Köstinger, A., et al.: Numerical simulation of various electrode configurations in impedance cardiography to identify aortic dissection. In: Proceedings of ICEBI (2019) 4. Gabriel, S., et al.: The dielectric properties of biological tissues. Phys. Med. Biol. 41(11), 2231 (1996)

Numerical Simulation of ICG Changes in Case of Chronic AD

59

5. Ulbrich, M., et al.: Influence of physiological sources on the impedance cardiogram analyzed using 4D FEM simulations. Physiol. Meas. 35, 1451–1468 (2014) 6. Alastruey, J., et al.: On the impact of modelling assumptions in multiscale, subject-specific models of aortic haemodynamics. J. R. Soc. Interface 13, 20160073 (2016) 7. Chen, D., et al.: A patient-specific study of Type-B aortic dissection: evaluation of true-false lumen blood exchange. Biomed. Eng. Online 12, 65 (2013) 8. Cheng, Z., et al.: Analysis of flow patterns in a patient-specific aortic dissection model. J. Biomech. Eng. 132(5), 051007 (2010). https://doi.org/10.1115/1.4000964

Analysis of Silicone Additives to Model the Dielectric Properties of Heart Tissue Leonie Korn(B) , Simon Lyra, Steffen Leonhardt, and Marian Walter Medical Information Technology, RWTH Aachen University, Pauwelsstr. 20, 52074 Aachen, Germany [email protected] https://www.medit.hia.rwth-aachen.de

Abstract. VADs (Ventricular Assist Devices) support the weakened heart by pumping blood carrying oxygen and life-essential nutrition into the organs. The advantageous location of minimal-invasive catheterbased VADs in the left ventricle and the aorta can be used to determine relevant cardiac parameters to achieve physiological and individual control of the VAD according to the needs of the patient. These parameters can be used to accelerate weaning from the pump, leading to an improvement in the patient’s quality of life. To date, impedance technology is a suitable tool for monitoring the volume of hollow organs and analysing electrical changes in tissue properties. Using this technique in the heart, we believe that cardiac recovery in VAD therapy can be improved by measuring left ventricular volume and myocardial property changes due to insufficient perfusion. In previous work we presented the development of silicone anatomical heart phantoms for impedance measurement in VAD therapy. In this work we introduce different silicone additives to model the electrical properties of heart tissue. Therefore, two additives, carbon and barium titanate, were analysed with an LCR-meter regarding amplitude and phase of impedance between 1 kHz and 100 kHz and compared to pure silicone samples. Additionally, electrical equivalent circuit models of the samples were investigated and a simulation of sample thickness was performed. We found that mixtures of carbon + silicone have resistive properties in contrast to mixtures of silicone + barium titanate which can be modelled by connecting a capacitor parallel to a resistor. The variation of sample thicknesses showed that electrical properties similar to those of heart tissue can be achieved with the presented additives.

Keywords: Silicone additives properties · VAD therapy

1

· Heart muscle · Dielectric tissue

Introduction

By 2015, more than 400 million people were suffering from heart disease. Due to an unhealthy lifestyle and genetic risk factors, heart disease is still the leading c Springer Nature Singapore Pte Ltd. 2020  P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 60–66, 2020. https://doi.org/10.1007/978-981-13-3498-6_10

Model of Dielectric Properties of Heart Tissue

61

cause of death in developing countries [1]. Inadequate blood circulation and oxygen supply lead to cardiogenic shock if the required metabolic requirements are not met. When treating cardiogenic shock, it is therefore essential to ensure sufficient blood flow to the organs and in particular to the coronary arteries. VADs are used to improve blood circulation and thus give the physician time to make further decisions. Until now, VADs run at a fixed speed and do not adapt to the individual needs of the patient. The so-called physiological control is a strategy aimed at controlling the physiological state of the patient and not just the device itself [2]. Therefore, suitable parameters for determining the patient’s condition must be assessed. The position of minimal-invasive pumps in the left ventricle and the aorta is perfect to integrate sensors on the surface of the VAD. For instance, pressure sensors have already been attached to the Impella (Abiomed Inc., Danvers, USA) VAD pump. To observe the electrical behaviour of the surrounding tissue, the placement of electrodes on the VAD is feasible. Baan et al. [3] have already shown that there is a correlation of catheter-based impedance measurement in the ventricle and its volume. Furthermore, Schaefer et al. [4] found that the dielectric properties of heart muscle tissue change during ischaemia. In order to improve the measurement technique and to obtain a suitable parameter to characterize the status of the heart, we are currently working on the development of a left ventricle phantom that represents the heart’s dielectric properties. In [5] we have already introduced the casting process of a left ventricle phantom and analysed the silicone samples with different concentrations and sizes of carbon additives with regard to their electrical properties. While manufacturing conductive samples is possible, capacitive behaviour as observed in biological tissues has not been integrated successfully in non-biological samples yet. In this work we present the analysis of two silicone additives, carbon and barium titanate, and their series connection in order to model the dielectric properties of heart muscle tissue.

2 2.1

Methods Silicone Samples

The experiments were carried out using silicone with a shore scale of A00 (Silikonfabrik.de, Ahrensburg, Germany). Samples were cast in a cylindrical shaped mould with a diameter of 39 mm. Two additives were used to model the properties of heart muscle: Carbon, which is a highly conductive material S ) and Barium titanate (BaTiO3 ), a mixed oxide of barium and (σC ≈ 62.5 · 103 m titanium. BaTiO3 belongs to the group of electro-ceramics and is frequently used in class 2 ceramic capacitors. Being a ferroelectric, it has a high relative permittivity in the range of r = 1200...10000. Compared to this, silicone elastomers have a relative permittivity of r = 3...9. Three different samples where cast: pure silicone, silicone with carbon and silicone with barium titanate. Generating the silicone + carbon-sample, silicone was mixed with 0.7 parts per hundred rubber (phr) of carbon cutlets with an

62

L. Korn et al.

average length of 3 mm and 1 phr of carbon black powder VULCANR XC72 from Cabot Corporation (Boston, USA). For the sample of silicone + BaTiO3 , 10 g of silicone and 6.37 g of BaTiO3 were used, which is equivalent to 63.7 phr of BaTiO3 . All samples have been vacuum treated and cured at room temperature (approx. 22 ◦ C). For sample analysis a precision LCR-Meter E4980A from Keysight (Santa Rosa, USA) was used. All samples were placed between two circular copper circuit boards in order to analyse their dielectric properties. A weight of 35.6 g was placed on top of the copper plates to ensure the same surface pressure on all samples. The measurements were conducted with all three samples (pure silicone, silicone + carbon and silicone + BaTiO3 ) and additionally with the series connection of silicone + carbon and silicone + BaTiO3 (hereafter referred to as carbon + BaTiO3 ). Basically, the idea of connecting the samples in-series was that carbon fibres do not short-circuit the BaTiO3 -particles and resistive and capacitive influences are maintained compared to a mixture of both additives in one sample. All measurements were analysed by their complex admittance in order to obtain the complex permittivity and the resulting conductivity (Eqs. 1 and 2) in a frequency range of 1 kHz to 100 kHz: Ymeas = jωC = 0

A ω(r + jr ) d

σ(ω) = 0 ωr

(1) (2)

in which A is the area of the electrodes and d the distance between them. ω is the angular frequency and 0 the permittivity of vacuum. Table 1 shows the measured thicknesses of all samples and their combination. The area of the electrodes was A = 0.0012 m2 . Table 1. Sample thickness Silicone

dmeas,S = 7 mm

BaTiO3

dmeas,B = 7 mm

Carbon

dmeas,C = 11 mm

Carbon + BaTiO3 dmeas,serial = 18 mm

2.2

Simulation and Analysis

A sophisticated analysis of the properties of the combination of silicone and the two materials was performed. Therefore, to enable heart tissue phantom development in the future we modelled the samples by their electrical equivalents. Assumptions were made on the basis of the measurements of the individual samples (plotted in Fig. 2, which are described in detail in the results section).

Model of Dielectric Properties of Heart Tissue

63

The silicone + carbon sample has a phase angle of almost zero degree whereas silicone + BaTiO3 has a phase angle around −89.4◦ over all frequencies. Compared to the pure silicone sample (phase of −90◦ ), the sample of silicone + BaTiO3 has a small additional resistive component. Therefore, the mixture of silicone + carbon is considered to be purely resistive and the mixture of silicone + BaTiO3 a parallel circuit of a resistor and a capacity (see Fig. 1).

Fig. 1. Assumed electrical equivalent circuits of the samples of silicone + carbon (top, left), silicone + BaTiO3 (top, right) and of the serial connection of both samples (bottom).

The series connection of one silicone + carbon sample and one silicone + BaTiO3 sample leads to the electrical equivalent circuit of Yserial (see Fig. 1, bottom). The resulting admittance of the series connection is then according to Eq. 3: YCarbon · YBaT iO3 ≈ Ymeas . (3) Yserial = YCarbon + YBaT iO3 The results of the simulated and measured admittance of the series connection are compared to verify the assumptions. For all calculations, the material properties (r and σ) were calculated to be independent of the dimensions of the sample. Having measured the dielectric properties of the individual samples, the conductivity and permittivity of different series-connected combinations of both materials can be simulated. For this reason, we have varied the thickness of the samples in the simulation. The overall sample thickness is set to 1 and we define the thickness of the sample consisting of BaTiO3 to be dB . It follows directly the sample thickness of carbon dC = 1 − dB (see also Fig. 3, left). Please note that for reasons of clarity, only the thickness of silicone + BaTiO3 is specified in the graphs.

3

Results

Figure 2 shows the measurements of all samples and the series combination of carbon + BaTiO3 . The complex impedance separated in magnitude and phase is shown on the left and the resulting dielectric properties of the mixtures are shown on the right. Additionally, both plots contain the simulated series connection

64

L. Korn et al.

Fig. 2. Left: Comparison of measurements over frequency separated in magnitude [Ω] and Phase [◦ ] including the calculated series combination of carbon + BaTiO3 . Note that in the Phase diagram all values except YCarbon are between −89.2◦ and −90.0◦ ; Right: Dielectric properties of the samples and the results from the calculated admittance of the series connection.

of carbon + BaTiO3 for comparison. The sample of silicone solely filled with carbon has a frequency independent magnitude of approx. 20 Ω and a phase S and its of almost zero degrees. The resulting conductivity is σCarbon = 0.45 m permittivity dependent on frequency Carbon = 150...50. The sample of pure silicone and the one filled with BaTiO3 behave like dielectrics. As mentioned before, they show capacitive behaviour and have a magnitude in the range of MΩs. Measuring the series combination of carbon + BaTiO3 (Fig. 2, right), it can be observed that the dielectric properties increase in comparison to pure silicone and that both permittivity and conductivity increase significantly in comparison to the other samples. Therefore, the assumption that the measured sample of carbon + BaTiO3 can be modelled by its electrical equivalent as a series connection of a resistor and the parallel circuit of a capacitor and a resistor (see Fig. 1, bottom) is a reasonable fit. This is evident from the results of the simulation of Yserial,simu . It can be seen that there is a small difference between the simulation Yserial,simu and the measurement Yserial . This may be due to measurement errors caused by varying contact pressure, inaccuracies in sample dimensions, or disregarded contact resistances. Based on the evaluation of the measured material properties, it was possible to simulate different sample thicknesses of the series combination. These results are depicted in Fig. 3. With a significant reduction of the sample size dB of the mixture of silicone + BaTiO3 , the dielectric properties increase to a conductivity S S and to a permittivity of approx.  = 4.5 · 104 m . of σ = 0.1 m

Model of Dielectric Properties of Heart Tissue

65

Fig. 3. Left: Principle of the simulated variation of sample thickness for carbon + BaTiO3 ; Right: Dielectric properties of the simulated sample thicknesses. All S . conductivities except dB = 0.0001 are smaller than 10−7 m

4

Conclusion and Discussion

This study was an approach for modelling the dielectric properties of passive S heart muscle tissue with a conductivity in the range of σheart = 0.05...0.35 m  3 5 and a permittivity heart = 10 ...10 [6]. The objective of this research is to produce a silicone phantom with similar properties of heart muscle tissue to simplify further research in the field of cardiac bioimpedance measurements. From the results shown in the previous section it can be seen that we can manipulate the dielectric properties of silicone by mixing in different additives. Carbon is used to gain resistive behaviour, while BaTiO3 is able to increase the permittivity of silicone. The series connection of these two samples increases both, conductivity and permittivity, but is far from being close to the properties of heart muscle tissue. We showed that assumptions for modelling the electrical equivalents of the samples are plausible, as measured and simulated results agree. The electrical equivalents allow the dielectric properties to be analysed based on sample thickness and thus simplifies further sample generation. Therefore, a simulation of sample thicknesses was performed. The results show that if we apply a very thin layer of silicone + BaTiO3 (dB = 0.0001) in series with a silicone + carbon sample (dC = 0.9999), dielectric properties similar to those of heart tissue could be reproduced. Further investigations must be carried out by physically casting samples within the ranges of the simulated sample thicknesses. This will be challenging due to the very thin layer that has to be applied to the complex shape of the ventricle phantom. If the simulation results can be reproduced, a left heart phantom with the specified concentrations can be cast. In addition, a wider frequency range for admittance measurements should be considered. In the future,

66

L. Korn et al.

we plan to develop a model that also takes into account the concentration of the additives in order to obtain more information for the generation of suitable material. We found that problems with homogenization and remaining air in the samples are still experienced as a result of the casting process. Therefore, this procedure needs to be further investigated and optimized. To summarize, the present study deals with the analysis of silicone additives due to their dielectric properties. For this purpose, the samples were modelled as electrical equivalents in order to obtain more possibilities for combination. The variation of sample thickness showed that the behaviour of the dielectric properties of the heart muscle tissue can be achieved. Acknowledgements. This work has been funded by the Federal Ministry of Education and Research (BMBF, Germany) and is part of the project inHeart (grant number 13GW0118C). Conflict of Interest. The authors declare that they have no conflict of interest.

References 1. Roth, G.A., Johnson, C., Abajobir, A., Abd-Allah, F., Abera, S.F., Abyu, G., Ahmed, M., Aksut, B., Alam, T., Alam, K., et al.: Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. J. Am. Coll. Cardiol. 70, 1–25 (2017) 2. AlOmari, A.H., Savkin, A.V., Stevens, M., Mason, D.G., Timms, D.L., Salamonsen, R.F., Lovell, N.H.: Developments in control systems for rotary left ventricular assist devices for heart failure patients: a review. Physiol. Meas. 34, R1 (2012) 3. Baan, J., Van Der Velde, E.T., De Bruin, H.G., Smeenk, G.J., Koops, J., Van Dijk, A.D., Temmerman, D., Senden, J., Buis, B.: Continuous measurement of left ventricular volume in animals and humans by conductance catheter. Circulation 70, 812–823 (1984) 4. Schaefer, M., Gross, W., Ackemann, J., Gebhard, M.M.: The complex dielectric spectrum of heart tissue during ischemia. Bioelectrochemistry 58, 171–180 (2002) 5. Korn, L., Lyra, S., R¨ uschen, D., Pugovkin, A., Telyshev, D., Leonhardt, S., Walter, M.: Heart phantom with electrical properties of heart muscle tissue. Curr. Dir. Biomed. Eng. 4, 97–100 (2018) 6. Gabriel, S., Lau, R.W., Gabriel, C.: The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz. Phys. Med. Biol. 41(11), 2251 (1996)

A Short Review of Membrane Models for Cells Electroporation Jéssica R. da Silva

, Raul Guedert(&) , Guilherme B. Pintarelli and Daniela O. H. Suzuki

,

Institute of Biomedical Engineering (IEB/UFSC), Federal University of Santa Catarina, Florianópolis, SC, Brazil [email protected]

Abstract. This article aims to present some bioimpedance models and mathematical solutions used to describe the theoretical process of electroporation in biological cells. Throughout this article three different models are mentioned, with a greater focus on the last two, which are focused on the formation of pores which theoretically influence the conductivity of the membrane. Keywords: Electroporation model


Bioimpedance
Cell membrane
Conductivity

1 Introduction Electroporation is a phenomenon that increases the permeability of cells and tissues due to pores creation on cellular membrane. The phenomenon occurs when the cells are exposed to sufficiently high electric field. There are three different types of occurrences depending on the characteristics of the applied electric field. Firstly, Reversible Electroporation (RE) occurs when the electric field is high enough to exceed the critical threshold for pores creation and the cells are still able to return to their initial state. Secondly, Irreversible Electroporation (IRE) occurs at higher electric field values. The excessive pore density induces the cell osmotic imbalance or homeostasis loss, resulting in cell death. Lastly, Joule heating may occur, mainly due to excessive exposing time [1]. The factor that determines whether electroporation will be reversible is the transmembrane potential Vm. Considering a theoretical cell of radius equal to 10 lm, when this potential rises within the range of 0.2 V  Vm  1 V, the pores open and close at the end of the stimulus, reestablishing the membrane. In contrast, in the case where Vm exceeds 1 V, the pores will remain open, causing ionic imbalance, membrane rupture and cell death [2, 3]. The membranes are phospholipid bilayers. The hydrophilic phosphates are at the outer ends; the lipids, which are hydrophobic, form the inner layer of the membranes. In addition, there are proteins that, when binding to ions, allow transport through the bilayer, forming ionic channels. A classical electroporation process in the cell membrane refers to the creation of aqueous pathways (pores) that cross the membrane

© Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 67–74, 2020. https://doi.org/10.1007/978-981-13-3498-6_11

68

J. R. da Silva et al.

through the lipid bilayer, when electric fields are applied. Thus, allowing molecules to pass through the membrane. The cell membrane is a dielectric and the pore opening causes a decrease of the electrical resistance of the membrane (i.e., facilitates ion transport). Thus, a way of analyzing the process of electroporation is to indirectly measure changes in electrical properties before, during and/or after the application of electroporation, independently of the level of treatment, in other words, in a single-cell [4], cell suspension [5, 6] and tissues [7, 8]. The analysis of electroporation during application of the protocols can be performed by acquisition of instantaneous apparent impedance (or impedance magnitude V/I). The Fig. 1 shows this type of situation, the impedance decreases in a non-linear or non-ohmic way as a function of the occurrence of electroporation, in other words, by increasing the amplitude/duration/number of pulses of the electroporation protocol. The decrease in apparent impedance causes an increase in the electric current [3, 7, 9].

Fig. 1. Representation of the apparent impedance change as a function of electroporation. If electroporation threshold is exceeded the apparent impedance V/I decreases.

Despite the use of the apparent impedance analysis, it is known that errors are introduced by dispersive effects. These effects are concentrated at lower frequencies than 10 kHz. Rectangular pulses have many frequency components. Therefore, the analysis of the changes during the pulses must be contaminated with dispersive effects, since it is not possible to separate them by this method [10, 11]. One way of reducing low frequencies dispersive effects is the use of protocols with high frequency components [10] or a combination of rectangular pulses with sinusoids above 1 kHz [12]. However, in electrochemotherapy, it is defaulted to use the European Standard Operating Procedures of Electrochemotherapy pulse duration of 100 ls [13, 14]. The comparative impedance spectrum analysis before and after the application (e.g., small signal analysis) provide data with lower contamination of dispersion effects (optimal frequency for cell and membrane sensing is 100 Hz to 100 kHz) [10, 15]. Nevertheless, this type of analysis is limited to changes after electroporation (i.e., mainly due to ionic leakage), in other words, it does not allow access to the state of the cells during the process. This information is desired on future systems with real-time monitoring and feedback of the electroporation status.

A Short Review of Membrane Models for Cells Electroporation

69

The observation of pores is often done indirectly since the pores have opening times and dimensions in the order of nanometer and nanoseconds respectively. The use of models helps in understanding the experimental results of electroporation. The analysis of single cell membrane can be extrapolated to cell suspension. This fact highlights the importance of cell membrane models for understanding experimental results with macroscopic bioimpedance [16]. The objective of this work was to list some existing model methods on the electroporation of cell membrane regarding membrane conductivity.

2 Electroporation Models Several mathematical models have been developed to describe changes in the cell membrane in the presence of an external electric field. The importance of mathematical modeling is both the possibility of understanding this microscopic phenomenon, but also expand the research in the area without excessive costs with experiments. 2.1

Kinetics Models

This approach uses four steps kinetics to describe the pore formation phenomenon [17]. The reaction occurs in cascade, accordingly as shown in the Eq. 1. In the first step (C), the lipids are closed and intact, in a condition of low permeability. The second stage (P1) is characterized by tilted lipid headgroups. The third stage is where the formation of pre-pores occurs (P2) and, the fourth stage (P3) the final pores are formed. The pore opening is mainly dependent of the electrical field amplitude applied to the cell which induces a Vm. C P1 P2 P3

ð1Þ

Coefficients (kx) determine the changing state of the membrane; they are called ‘pore formation coefficients’ (kp, k1, k2, k3) and ‘pore closure’ (k−1, k−2, k−3). Pore formation coefficients are f( Vm2 ). The reaction is described by a four-equation differential equation system (equations not shown). The membrane conductivity is according to Eq. 2. r; tÞ ¼ rm0 þ ½P1 ð~ r; tÞ  rp; 1 þ ½P2 ð~ r; tÞ  rp; 2 rm ð~

ð2Þ

The vector ^r represents the membrane position, which is an angle value between 0º to 360º, and t represents time. The scalars P1ð^r; tÞ and P2ð^r; tÞ represent the normalized distribution of each membrane lipid state relative to the initial value of the closed state C ð^r; 0Þ. The rm0 is the initial conductivity of membrane. The intrinsic pore conductivities rp;1 (P1) and rp;2 (P2) are f(Vm2 ).

70

2.2

J. R. da Silva et al.

Asymptotic Model

The induced voltage on the membrane of a spherical cell of radius R, inserted into a uniform electric field E0 can be described from Eq. 3 based on the Laplace equation in the time domain, valid for a thin dielectric shell immersed in a conductive environment. The equation of the transmembrane potential is given by [6], in which 1,5 is a factor that reflects the electrical and geometric properties of the cell when it is assumed that the conductivity of the membrane is much smaller than the conductivities of the cytoplasm and aqueous solution, moreover, the cell is considered approximately spherical.
 t Vm ¼ 1;5  E0  R  cosh 1  esm

ð3Þ

h is the angle formed between the electric field vector applied and the vector normal to the membrane surface, sm is the plasma membrane loading time, and t is the time of the membrane to the applied electric field. The pore density and the electrical conductivity are obtained from a non-linear mathematical model, resulting from the simplification of the partial differential equation of Smoluchowski [18]. By using a simplified version, the dynamics of the electroporation process can be studied through its asymptotic model. According to the simplification, the model can be described by an ordinary differential equation with lower computational cost. A complete model for the study of the pore density formed during the electroporation process considers the formation of pores with different radius sizes. The rate of pore creation can be observed in Eq. 4, in which a is the coefficient of pore formation rate, Vm(t) is the transmembrane potential, Vep is the characteristic voltage of electroporation, N0 is pore density at equilibrium and q is the coefficient of electroporation: dN ðtÞ ¼ ae dt




Vm ðtÞ Vep

2 0

q @1  N ð t Þ e No




Vm ðtÞ Vep

2 1 A

ð4Þ

R1 The pore density is given by N ðtÞ ¼ r nðr; tÞ dr, where r is the initial radius of hydrophilic pore. The conductivity rpm inside the pore in the membrane is given by Eq. 5 in which re represents the conductivity of the extracellular solution likewise rc is the cytoplasm conductivity. rpm ¼

re  rc
 ln rrec

ð5Þ

The conductivity value is required for the calculation of the average conductivity of the membrane in the regions where the pores are formed according to Eq. 6, in which rm0

A Short Review of Membrane Models for Cells Electroporation

71

is the conductivity of the membrane prior to electroporation, rp is the conductivity in the middle of the pore and rp is the pore radius. rm ðtÞ ¼ rm0 þ N ðtÞrp prp2 K K¼

evm  1 w0 mvm evm ww0 e0 mvm

w0 þ mvm

 w0 e wo þ mvþm mvm

ð6Þ ð7Þ

The term K is a coefficient dependent on vm, which is the dimensionless transmembrane potential vm ¼ Vm kTq and is obtained by Eq. 7 in which w0 is the energy barrier within the pore and t is the relative depth of the pores in the membrane. 2.3

A Theoretical Study of a Single-Cell Electroporation in a Microchannel

This study aims to mathematically modeling the electroporation process using the Asymptotic Model in order to identify areas where nanopore formation predominates [19]. The numerical results obtained demonstrate the relationship between the frequency and the generation rate of nanopores, as well as the dependence between the variation of time and the shape and number of pores created. In order to find the electric field inside Ui and outside Ue of the cell, the Laplace equation is required, as shown in Eq. 8. r2 /i;e ¼ 0

ð8Þ

n
^J ¼ 0

ð9Þ

n:ðsi r/i Þ ¼ n:ðse r/e Þ

ð10Þ

n:ðsi r/i Þ ¼ cm

@Vm þ g1 ðVm  Vrest Þ þ Ip @t

ð11Þ

According to this theory, the electric current density, which can be described in Eq. 11, should be continuous across the membrane and it is formed by capacitive m current (cm @V @t ), the current through the protein channels (g1 ðVm  Vrest Þ) and the current through the created nanopores (Ip ). Ip can be found as the Eq. 12 demonstrates. Ip ðt Þ ¼

 1 Xm  i r ; Vm j¼1 p j DA

ð12Þ

Here, m is the number of created nanopores and ip is the current through each nanopore. Vm is the transmembrane potential and is defined as: Vm ¼ Ui ðt; a; hÞ  Ue ðt; a; hÞ

ð13Þ

72

J. R. da Silva et al.

By applying the electric field, nanopores start to develop in order to minimize the energy of the membrane. The rate of change the radius of the nanopores can be   dr determined by dtj ¼ U rj ; Vm; reff and Eq. 14, in which reff is the effective tension of the membrane. 

U rj ; Vm ; reff



D ¼ kT

!   4 r 1 Vm2 FMax 4b  2pc þ 2preff r þ r r 1 þ r þrh rt

ð14Þ

This study demonstrates that the pore distribution does not occur uniformly in the membrane. As the induced transmembrane potential is concentrated at the poles (where the cell membrane is nearest to the electrodes), the biggest pores are also concentrated at the poles and nanopore population expands toward the equator.

3 Discussion The choice of the appropriate model depends on what to evaluate. The kinetic model can predict all electrical and diffusion parameters for each type of pulse. This model is useful when electrical and diffusion properties of the cell to be electroporated is already known, allowing the optimization of some protocols. The asymptotic model allows us to investigate both spatial and temporal aspects of the pore formation process. The third model presented here focuses on the influence of magnitude and duration of the pulse, the position of the electrodes and height of the micro-channel in the permeability process.

4 Conclusions Numerical modeling is a regular approach in evaluation of electroporation and can support numerous areas of medicine and biotechnology. The importance of theoretical studies involving mathematical modeling and simulators is mostly based on the requirement of less resources than would be necessary for an experimental investigation and the possibility of assuming abnormal or extreme regimes of operation of a given process. In addition, the use of mathematical models has been favored by the development of new technologies, software and calculation methods, which optimize the search process for results. Acknowledgements. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. We would like to thank the Brazilian research funding agencies CAPES and CNPq for the scholarships granted to the postgraduate students. Conflict of Interest. The authors declare that they have no conflict of interest.

A Short Review of Membrane Models for Cells Electroporation

73

References 1. Davalos, R.V., Mir, L.M., Rubinsky, B.: Tissue ablation with irreversible electroporation. Ann. Biomed. Eng. 33(2), 223 (2005) 2. Kotnik, T., Kramar, P., Pucihar, G., Miklavcic, D., Tarek, M.: Cell membrane electroporation - Part 1: The phenomenon. IEEE Electr. Insul. Mag. 28(5), 14–23 (2012) 3. Suzuki, D., Yamada, A., Amano, T., Yasuhara, R., Kimura, A., Sakahara, M., Tsumaki, N., Takeda, S., Tamura, M., Nakamura, M., Wada, N.: Essential mesenchymal role of small GTPase Rac1 in interdigital programmed cell death during limb development. Dev. Biol. 335(2), 396–406 (2009) 4. Huang, Y., Rubinsky, B.: Microfabricated electroporation chip for single cell membrane permeabilization. Sens. Actuators A: Phys. 89(3), 242–249 (2001) 5. Pavlin, M., Leben, V., Miklavčič, D.: Electroporation in dense cell suspension—theoretical and experimental analysis of ion diffusion and cell permeabilization. Biochim. Biophys. Acta (BBA) Gen. Subj. 1770(1), 12–23 (2007) 6. Suzuki, D.O., Ramos, A., Ribeiro, M.C., Cazarolli, L.H., Silva, F.R., Leite, L.D., Marques, J.L.: Theoretical and experimental analysis of electroporated membrane conductance in cell suspension. IEEE Trans. Biomed. Eng. 58(12), 3310–3318 (2010) 7. Neal II, R.E., Garcia, P.A., Robertson, J.L., Davalos, R.V.: Experimental characterization and numerical modeling of tissue electrical conductivity during pulsed electric fields for irreversible electroporation treatment planning. IEEE Trans. Biomed. Eng. 59(4), 1076–1085 (2012) 8. Ramos, L.C., Pintarelli, G.B., Altenhofen, D., Suzuki, D.O.H.: Micro electropermeabilization system for cell medium conductivity change measurement of erythrocytes cells. In: 1st World Congress on Electroporation and Pulsed Electric Fields in Biology, Medicine and Food and Environmental Technologies, pp. 87–90. Springer, Singapore (2016) 9. Langus, J., Kranjc, M., Kos, B., Šuštar, T., Miklavčič, D.: Dynamic finite-element model for efficient modelling of electric currents in electroporated tissue. Sci. Rep. 6, 26409 (2016) 10. Bhonsle, S.P., Arena, C.B., Sweeney, D.C., Davalos, R.V.: Mitigation of impedance changes due to electroporation therapy using bursts of high-frequency bipolar pulses. Biomed. Eng. Online 14(3), S3 (2015) 11. Grimnes, S., Martinsen, Ø.G.: Sources of error in tetrapolar impedance measurements on biomaterials and other ionic conductors. J. Phys. D Appl. Phys. 40(1), 9 (2006) 12. Ramos, A., Schneider, A.L., Suzuki, D.O., Marques, J.: Sinusoidal signal analysis of electroporation in biological cells. IEEE Trans. Biomed. Eng. 59(10), 29 (2012) 13. Marty, M., Sersa, G., Garbay, J.R., Gehl, J., Collins, C.G., Snoj, M., Billard, V., Geertsen, P. F., Larkin, J.O., Miklavcic, D., Pavlovic, I.: Electrochemotherapy – an easy, highly effective and safe treatment of cutaneous and subcutaneous metastases: results of ESOPE (European Standard Operating Procedures of Electrochemotherapy) study. Eur. J. Cancer Suppl. 4(11), 3–13 (2006) 14. Gehl, J., Sersa, G., Matthiessen, L.W., Muir, T., Soden, D., Occhini, A., Quaglino, P., Curatolo, P., Campana, L.G., Kunte, C., Clover, A.J.P.: Updated standard operating procedures for electrochemotherapy of cutaneous tumours and skin metastases. Acta Oncol. 57(7), 874–882 (2018) 15. Bürgel, S.C., Escobedo, C., Haandbæk, N., Hierlemann, A.: On-chip electroporation and impedance spectroscopy of single-cells. Sens. Actuators B: Chem. 210, 82–90 (2015)

74

J. R. da Silva et al.

16. Pavlin, M., Miklavčič, D.: Effective conductivity of a suspension of permeabilized cells: a theoretical analysis. Biophys. J. 85(2), 719–729 (2003) 17. Neumann, E., Tönsing, K., Kakorin, S., Budde, P., Frey, J.: Mechanism of electroporative dye uptake by mouse B cells. Biophys. J. 74(1), 98–108 (1998) 18. Neu, J.C., Krassowska, W.: Asymptotic model of electroporation. Phys. Rev. E 59(3), 3471 (1999) 19. Movahed, S., Li, D.: A theoretical study of single-cell electroporation in a microchannel. J. Membr. Biol. 246, 151–160 (2013). https://doi.org/10.1007/s00232-012-9515-6

Body Composition

Data Views Technology of Bioimpedance Vector Analysis of Human Body Composition Svetlana P. Shchelykalina1,2(&), Dmitry V. Nikolaev1, Vladimir A. Kolesnikov1, Kontantin A. Korostylev1, and Olga A. Starunova3 “SRC Medas”, 2nd Baumanskaya, 7 p. 1a, Moscow, Russia [email protected] 2 Pirogov Russian National Research Medical University (RNRMU), Moscow, Russia Central Public Health Research Institute of the Ministry of Health of Russia, Moscow, Russia 1

3

Abstract. In 1994, A. Piccoli et al. proposed BIVA—an alternative form of BIA data presentation. Standard values of BIVA are usually presented as 75% and 95% tolerance ellipse. Aim of the work: to assess the accuracy of the classical bioimpedance vector analysis for the population of Russia and to develop the option of a twodimensional representation of the data of bioimpedance analysis of human body composition. Materials and methods The present study used data from 1,635,891 patients aged 5 to 85 who underwent bioimpedance research as part of visiting Russian health centers in 2009–2015. The BIVA ellipses for each year of life were constructed, and their congruence with each other and the ellipses from A. Piccoli’s work using the conics by Bernard Desgraupes’ package in an R environment. Animation software was also developed, which, based on raw data, constructed slices of the actual two-dimensional distribution of any selected pairs of bioimpedance human body composition parameters. Results and discussion Calculated according to the Russian data, the BIVA tolerance ellipses differed from those in A. Piccoli’s work. The ellipses of the Russian population had a non-concentric location: the centers of 50%, 75% and 95% tolerance ellipses do not coincide. In many ages, not only the center displacement was detected, but also changes in the angle of inclination of the main axis of 95% and 50% tolerance ellipses. Ellipses for adjacent ages were different too. The slices of the actual two-dimensional distribution had an irregular shape that varied greatly with age, especially for 95% tolerance cloud. Conclusion BIVA ellipses of the Russian population showed a big difference from Piccoli’s. For an adequate assessment and minimization of possible errors, we should use localized reference values for each gender and age. The proposed two-dimensional representations allow us to analyze four pairs of BIA parameters. © Springer Nature Singapore Pte Ltd. 2020 P. Bertemes-Filho (Ed.): ICEBI 2019, IFMBE Proceedings 72, pp. 77–83, 2020. https://doi.org/10.1007/978-981-13-3498-6_12

78

S. P. Shchelykalina et al. Keywords: Bioimpedance analysis of human body composition
Bioimpedance vector analysis
Centile estimates
Graphical representations of data

1 Introduction For more than 30 years of practical application of bioimpedance analysis of human body composition (BIA), the BIA data presentation forms have been improved and developed: the protocols were repeatedly supplemented with new parameters, the ranges of normal values were refined, the age range of the subjects was expanded, and centile evaluations were performed. Figure 1 shows the main stages of the BIA protocol evolution. Protocols of rapid analysis, which are usually called the primary BIA protocols, contain one-dimensional graphical scales with a selected area of normal values. Such a presentation of the primary information of body composition is characteristic of the

Fig. 1. Main stages of the BIA protocols evolution

Data Views Technology of BIVA of Human Body Composition

79

absolute majority of models of body composition analyzers currently produced. The boundaries of the area of normal values for each parameter were repeatedly revised, as a result of obtaining more and more representative samples of experimentally obtained data and localized for ethnic populations. The main trend in constructed BIA protocols is to increase the information saturation of the graphical representations for convenience and to increase the speed of data perception by a doctor [1]. The nomenclature of the parameters used in the BIA has somewhat expanded with respect to the set of parameters available in the analyzers of the 80–90s. Active cell mass (ACM), skeletal muscle mass (SMM) and mineral mass (MM) were added to the indicators of fat mass (FM), fat-free mass, total, extra and intracellular water (TBW, ECW and ICW, respectively). In addition to the actual component of the body composition, the protocols of modern devices contain the values of metabolic correlates: basal metabolic rate (BMR), and the phase angle (PA) [2]. In the early 2000s, many countries obtained the results of population surveys by the BIA method. The data obtained in the Russian health centers, in the form of a sample of 808 thousands surveyed in 2010–2012 formed the basis for centile representations of body composition parameters of the Russian population [3]. In 1994, Piccoli et al. [4] proposed an alternative form of BIA data representation, later called bioimpedance vector analysis (BIVA). This technique uses measurement data at only one frequency—50 kHz. In BIVA, only two bioimpedance parameters are considered: resistance (R) and reactance (Xc), normalized to body length (BL). The data of each measurement is displayed by points on the plane in the coordinates of R/BL—horizontally and Xc/BL—vertically. The results are compared with the population data presented in the form of a system of nested tolerance ellipses. It is shown that in this coordinate system, 50%, 75%, and 95% tolerance ellipses remain almost unchanged in the age range from 18 to 55 and can be used to analyze data from the adult population [5]. The BIVA methodology is recommended to use for monitoring of dynamic changes [6]. Aim of the Work: To assess the accuracy of the classical bioimpedance vector analysis for the population of Russia and to develop the option of a two-dimensional representation of the data of bioimpedance analysis of human body composition.

2 Materials and Methods The present study used data from 1,635,891 patients aged 5 to 85 who underwent bioimpedance research as part of visiting Russian health centers in 2009–2015. The survey results are accumulated in the Federal Information Resource of Health Centers. In 2014, these data are available for analysis under grant No. 14-15-01085 of the Russian Science Foundation. Descriptive statistics of the data are presented in the form of medians and quantiles in Table 1.

80

S. P. Shchelykalina et al. Table 1. Median and quartile values of the main bioimpedance parameters used Sex, N Men 548,696 Women 1,087,195 Age, years 20.7 [13.2; 45.5] 42.5 [20.6; 57.3] Height, cm 170 [157; 177] 161 [156; 166] Weight, kg 68.2 [50.0; 80.0] 66.0 [55.0; 78.0] 2 25.0 [21.1; 29.9] BMI, kg/m 23.0 [19.3; 26.7] R50, Ohm 550.2 [483.3;636.6] 582.2 [518.7; 655.6] Xc50, Ohm 64.5 [57.0; 72.3] 65.7 [57.2; 66.6] Phi50, deg. 6.53 [5.92; 7.29] 6.35 [5.84; 6.93] FM, % 22.0 [16.2; 28.1] 32.3 [25.3; 38.8] SMM, % 51.6 [48.3; 55.4] 44.2 [40.4; 47.6] TBW, l 39.2 [28.7; 44.5] 32.6 [29.4; 35.8] ACM, kg 30.1 [21.1; 35.6] 24.7 [21.9; 27.6] BMI – body mass index, R50 – 50 kHz resistance, Xc50 – 50 kHz reactance, Phi50 – 50 kHz phase angle, FM – fat mass, percent from weight, SMM – skeletal muscle mass, percent from fat-free mass, TBW – total body water, ACM – active cell mass.

Data analysis included the construction of BIVA ellipses for each year of life and the assessment of their congruence with each other and those in A. Piccoli’s work. Since the observed distribution of parameters did not correspond to the Gaussian distribution law, then the least squares method was used for the construction. Data for construction were selected on the basis of the actual distribution density so as to form an interval between 45 and 55 centiles for constructing 50% tolerance ellipse, 72.5 and 77.5 for 75% tolerance ellipse, 94.5 and 95.5 for 95% tolerance. The analysis was performed using statistical programming in an R Studio environment using the package “conics” by Bernard Desgraupes package (University of Paris Ouest - Nanterre, Lab Modal’X (EA 3454)). Animation software (“SRC Medas”, 2016) was also developed, which, based on raw data, built sections of the actual two-dimensional distribution of any selected pairs of bioimpedance body composition assessment parameters.

3 Results and Discussion Calculated according to the Russian data, the BIVA tolerance ellipses at all ages significantly differed from those described in the work of Piccoli et al. [4] both in the position of the centers, and in ellipticity: The BIVA tolerance ellipses of the Russian population turned out to be more compressed and had a smaller area over the age of 18. The area ratio of 75% tolerance ellipses ranged from 0.35 to 0.52 in women and 0.71 to 1.0 in men. For 95% tolerance ellipses, these values ranged within 0.86–1.27 for women and 0.3–0.8 for men. Ellipses of early childhood ages were significantly shifted to the right and upward relative to the reference data [4, 5]. The area of Russian and the reference data [5]

Data Views Technology of BIVA of Human Body Composition

81

Fig. 2. 50%, 75% and 95% tolerance ellipses for 35 (solid line) and 36 (dashed line) years old men (A) and 60 (solid line) and 61 (dashed line) years old women (B). Points are 95% tolerance ellipses centers.

ellipses intersection fluctuated within 50–99%, depending on age, sex, and centile tolerance. The study of age-related changes in the ellipses of the Russian population revealed that the variability with age ranges from 20–35% per year in childhood to 0–15% in adulthood. The most pronounced fluctuations were observed in 95% tolerance ellipses. In Fig. 2, 50%, 75% and 95% tolerance ellipses for 35 and 36 year-old men (Fig. 2A) and 60 and 61 year-old women (Fig. 2B) are presented. The dashed line reproduces ellipses of older ages. The non-concentric arrangement of the ellipses of the Russian population stands out: the centers of 50%, 75% and 95% tolerance ellipses do not coincide (in Fig. 2 only 95% tolerance ellipses centers are shown). The same trend was present for all ages. The largest deviations from the position of the 50% tolerance ellipse in many ages were found to be the 95% tolerance ellipse: not only a center shift was observed, but also a change in the angle of inclination of the main axis. This phenomenon is due to the large variability of rarely-occurring combinations of initial parameters in the raw data and, as a result, in general, the form of data dispersion which is far from elliptical. Thus, use to describe the entire population of elliptical curves does not appear to be sufficiently reliable. The volume of bioimpedance research data collected at Russian health centers allows us to construct a vector representation of two-dimensional distributions without using a pre-selected curve shape. The effectiveness of the doctor’s work with the BIA data depends on the possibilities of a one-time graphical representation, as a rule, of 5–7 body composition parameters. The primary protocols use on average 12 one-dimensional, as a rule, graphic scales characterizing the patient’s body composition at the time of the study. The presentation of patient data as a point or a set of points on a population centile picture contains additional information about how typical such parameter values are in a given population. The way of presenting data in the classical BIVA protocol allows one to demonstrate the results of two parameters, indirectly interpreted in terms of the body composition without the scale of the time axis. At the same time, the very idea of BIVA can now be rethought as a bioimpedance vector analysis of human body

82

S. P. Shchelykalina et al.

Fig. 3. An example of visualization of bioimpedance research by 8 parameters, with centile lines, arranged in two-dimensional graphs. The points correspond to the process of reducing the fat mass in a patient without recorded associated diseases. The arrow points to the point of the first study.

composition—BIVA HBC. As a basis for the vector representation, body composition parameters can be used that are much more understandable to most doctors. Data visualization, containing more than two dimensions, is somewhat difficult to construct, and, more significantly, in interpreting the information. It is proposed to place four two-dimensional images of eight basic body composition parameters on the background of two-dimensional centile curves on one screen and on one sheet of the protocol. The technique of constructing two-dimensional curves of bioimpedance vector analysis of human body composition was considered in work [7]. Figure 3 shows an example of such a synthetic image consisting of four BIVA HBC images. The following pairs of main bioimpedance parameters were selected: R and Xc, SMM in percent and phase angle, total body water and BMI, active cell mass and FM in percent. The BIVA HBC centiles curves (Fig. 3) were constructed for 52–54-year-old women of the domestic population based on more than 64 thousand studies in health centers using Animation software (“SRC Medas”, 2016). The data points of the bioimpedance study of a specific patient are connected by straight lines. The figure shows that in the process of reducing the fat mass, the parameter values moved closer to the central zone, which corresponds to the most typical values for the Russian population. It is seen that two-dimensional centile curves are closed, of irregular shape. The points that form the chain of patient research data form a broken line. In the prevailing number of cases, the general direction of the broken line shows the trends of changes typical of a decrease in the fat mass values in a practically healthy patient.

Data Views Technology of BIVA of Human Body Composition

83

4 Conclusion Even at the stage of primary diagnosis of body composition, and at all stages of monitoring the treatment effectiveness, visual information for a comprehensive and complete assessment should be presented to the doctor in a convenient form for quick perception and understanding. For an adequate assessment and minimization of possible errors, we should use localized reference values for each sex and age. The proposed two-dimensional representations allow us to analyze four pairs of BIA parameters. Additionally, in the same protocols, the boundaries of the normal value zones corresponding to the usual BIA express analysis protocols can be demonstrated. The analysis of the correspondence of the BIVA ellipse parameters obtained from the Russian population data of different ages and those described earlier by A. Piccoli showed differences. Acknowledgment. This work was supported by grant No. 14-15-01085 of the Russian Science Foundation (Hwad: V. Starodubov). Conflict of Interest. The authors declare that they have no conflict of interest.

References 1. Tufte, E.R., Goeler, N.H., Benson, R.: Envisioning Information, vol. 126. Graphics Press, Cheshire (1990) 2. Nikolaev, D.V., Smirnov, A.V., Bobrinskaya, I.G., et al.: Bioimpedance Analysis of the Composition of the Human Body. Science Publishing, Moscow (2009) 3. Rudnev, S.G., Soboleva, N.P., Sterlikov, S.A., et al.: Bioimpedance study of body composition in the Russian population. RIO TsNIIOIZ, Moscow (2014) 4. Piccoli, A., Rossi, B., Pillon, L., et al.: A new method for monitoring body fluid variation by bioimpedance analysis: the RXc graph. Kidney Int. 46, 534–539 (1994) 5. Piccoli, A., Pillon, L., Dumler, F.: Impedance vector distribution by sex, race, body mass index, and age in the United States: standard reference intervals as bivariate Z scores. Nutrition 18(2), 153–167 (2002). https://doi.org/10.1016/S0899-9007(01)00665-7 6. Siváková, D., Vondrová, D., Valkovič, P., et al.: Bioelectrical Impedance Vector Analysis (BIVA) in Slovak population: application in a clinical sample. Open Life Sci. 8(11), 1094– 1101 (2013) 7. Nikolaev, D.V., Shchelykalina, S.P., Kolesnikov, V.A., et al.: Analysis of two-dimensional representations of age and gender distributions of body composition parameters of the population of the Russian Federation. In: ISS “SI-2016” Proceedings, Yoshkar-Ola, Russia, pp. 155–162 (2016)

Analysis of Electrical Bioimpedance for the Diagnosis of Sarcopenia and Estimation of Its Prevalence Clara Helena Gonzalez-Correa1(&), Maria Camila Pineda-Zuluaga2, and Luz Elena Sepulveda-Gallego1 1

2

Department of Basic Sciences for Health, Universidad de Caldas, Calle 65 No. 26-10, Manizales, Colombia [email protected] Department of Public Health, Universidad de Caldas, Manizales, Colombia

Abstract. Sarcopenia in older adults has become a public health problem associated with several adverse health outcomes that lead to high costs of care. This is why an accurate and timely identification of this condition is required, in order to carry out prevention and early intervention to reduce its prevalence. The most important components of sarcopenia are the loss of quantity and quality of skeletal muscle and diagnosis begins with the evaluation of these two dimensions. Whilst in developed countries there are cumbersome and expensive methods to assess the amount of muscle, such as dual energy x-ray absorptiometry (DEXA), magnetic resonance imaging (MRI) and computerized tomography (CT) countries with fewer resources are able to take advantage of less expensive and easy to apply techniques for the diagnosis of sarcopenia, such as the analysis of electrical bioimpedance. In this study, the latter method was used to estimate the amount of muscle mass, in conjunction with the assessment of the grip strength of the hand and the short battery of physical performance, in order to estimate the prevalence of sarcopenia in 210 older adults in Manizales, Colombia. For this, cut-off points obtained from 10 indicated normal physical performance [9, 10]. 2.9

Statistical Analyses

A descriptive analysis was performed using absolute and relative frequencies for qualitative variables; and mean and standard deviation for quantitative variables. Comparison of variables was performed using a bivariate analysis with the statistic x2. The analysis of the ROC curve was used to determine the predictive capacity of the phase angle. A p-value of