Biomedical Signal Processing: From a Conceptual ... - IEEE Xplore

SCANNING THE ISSUE

Biomedical Signal Processing: From a Conceptual Framework to Clinical Applications By M A T H I A S B A U M E R T , S e n i o r M e m b e r , I E E E ALBERTO PORTA, Member, IEEE ANDRZEJ CICHOCKI, Fellow, IEEE

T

oday, high fidelity data acquisition systems are able to record vast amounts of biomedical signals of electric and nonelectric origin invasively or noninvasively over extended periods of time, providing unprecedented opportunity to study the human body in health and disease. Biomedical signals have been widely used in the clinical setting for diagnostics, guiding therapy, patient monitoring, disease prevention, and risk assessment. With the more recent developments This special issue covers of health gadgets, various personal relevant contemporary biomedical data are now also easily challenges in the field of accessible to consumers and recorded biomedical signal virtually throughout all aspects of life. processing and Processing and interpreting these possibilities for future data provides a multifaceted set of challenges, including coping with technological nonstationary behaviors ubiquitously development. present in the time course of biomedical variables, separating sources from a mixture of signals typically observed from the body surface, detecting the often weak coupling between physiological processes in noisy measurement environments, extracting and classifying significant dynamical features, modeling the underlying physiological systems, understanding the cause–effect relation between interacting subsystems, and converting methodological parameters into relevant information able to drive the clinical process and produce a measurable impact on the health care system.

Digital Object Identifier: 10.1109/JPROC.2015.2511359

The present issue comprises 12 reviews that address the aforementioned challenges in applicative biomedical contexts by emphasizing the relevance of the methodological problem, the clinical importance of extracted information, and the possibility for future technological developments. The first three contributions deal explicitly with a fundamental issue in the analysis and interpretation of biomedical signals: the inherently nonstationary nature of biological processes especially when recorded over a long time. Clemson et al. [1] review the framework of chronotaxic systems for the reconstruction of time-dependent information to provide tools for studying signal dynamics and long-range correlations over multiple time scales; Nakamura et al. [2] review multiscale analysis methods and applications to biomedical data derived from wearable and/or biomedical sensing technologies over long periods of time; Leistritz et al. [3] describe strategies utilizing time-variant stochastic process models to extract diagnostically relevant information from brain

0018-9219 Ó 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

220

Proceedings of the IEEE | Vol. 104, No. 2, February 2016

Scanning the Issue

signals, exemplified by fetal EEG monitoring, EEG-based brain connectivity analysis, and functional MRI. Given the multitude and complexity of processes involved in and governing the human body, integrated analysis of multidimensional data is of utmost importance to advance our understanding of integrated human physiology and pathological variations. Identifying causality between intertwined biological processes based on data recorded in free-running, closedloop conditions is a crucial step for disentangling relationships between physiological variables. In this regard, Granger’s famous causality test for determining whether one time series provide unique information about the future evolution of another is gaining increasing importance in physiology research. Porta and Faes [4] review the Wiener–Granger causality paradigm with the specific aim of assessing causal interactions among components forming a network and working according to the principles of integration and segregation. Biomedical recordings frequently represent the spatio–temporal superposition of the activity distributed across different tissues and organs, obtained with sensors at suboptimal spatial and/or temporal resolution, e.g., positioned on the surface of the body or even at distance. Identifying independent sources from recordings of activity made with devices sensing the collective behavior is therefore a major challenge in interpreting multidimensional data. Zhou et al. [5] discuss component analysis approaches to tackle this matter. Brain–computer interfaces, allowing the control of various applications via thoughts, have received increasing REFERENCES [1] P. Clemson, G. Lancaster, and A. Stefanovska, “Reconstructing time-dependent dynamics,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015.2491262. [2] T. Nakamura, K. Kiyono, H. Wendt, P. Abry, and Y. Yamamoto, “Multiscale analysis of intensive longitudinal biomedical signals and its clinical applications,” Proc.

interest by the medical community during recent years as they can enable communication with locked-in syndrome patients and hold promise for motor function rehabilitation in people after brain stroke or controlling exoskeletons. In terms of signal processing, feature extraction and classification are the main tasks involved in brain–computer interfaces. Li et al. [6] review multimodal approaches to improve target detection and control of these systems. Aside from brain signals, understanding and interpreting signals from end-organs, i.e., skeletal muscles (called electromyograms) enables the design of prosthetic devices that can be controlled by amputees, for example, by activating muscles in stumps. Recent advances in the characterization of human motor units from surface electromyograms, using blind source separation techniques to identify the discharge times of individual motor units, are summarized by Farina and Holobar [7]. Technologies and signal processing algorithms for recording and decoding for neural prostheses that exploit peripheral nerve signals and electrocorticograms (ECoG) to interpret human intent and control prosthetic arms are reviewed by Warren et al. [8]. The electrocardiogram (ECG) is a well established, easily accessible, routinely acquired, and widely used signal for assessing the cardiac conduction process and diagnosing heart rhythm disorders. Laguna et al. [9] discuss more recent signal processing advances in capturing subtle variations in the ECG that have been associated with cardiac death. Among the heart rhythm disorders, atrial fibrillation is the most common arrhythmia, with a prevalence of

epidemic proportions. Catheterbased ablation of atrial sites that sustain the arrhythmia is an effective strategy of treatment, but identification of target sites remains challenging. Baumert et al. [10] review quantitative electrogram-based methods for guiding catheter ablation in atrial fibrillation with a special focus on how signal processing can be fruitfully exploited to improve practical clinically relevant procedures. Hosokawa and Sunagawa [11] have developed a closed-loop neuromodulation technology for baroreflex regulation that mimics natural blood pressure control. Dealing with the issue of “big data” in the clinical environment and its support to the decision making process, machine learning, and decision support in critical care is reviewed by Johnson et al. [12], focusing on issues of data corruption, compartmentalization, and complexity in regard to preprocessing of large volumes of biomedical signals from critically ill patients. We hope that the topics covered in this issue highlight relevant contemporary challenges in the biomedical field that, when successfully tacked, might considerably advance our understanding of physiological processes, our capability of treating pathological states, and our potential for interacting with human beings and the external environment. We advocate the systematic application of state-of-the-art techniques summarized in this issue. Even in fields outside the biomedical area, the reviewed signal processing methods may serve as a reference or first attempt for testing the performance of new, originally developed techniques whenever analogous challenges are faced. h

IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015.2491979. [3] L. Leistritz, K. Schiecke, L. Astolfi, and H. Witte, “Time-variant modeling of brain processes,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015. 2497144. [4] A. Porta and L. Faes, “Wiener–Granger causality in network physiology with

applications to cardiovascular control and neuroscience,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2476824. [5] G. Zhou et al., “Linked component analysis from matrices to high-order tensors: Applications to biomedical data,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2474704.

Vol. 104, No. 2, February 2016 | Proceedings of the IEEE

221

Scanning the Issue

[6] Y. Li et al., “Multimodal BCIs: Target detection, multidimensional control, and awareness evaluation in patients with disorder of consciousness,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2469106. [7] D. Farina and A. Holobar, “Characterization of human motor units from surface EMG decomposition,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2498665.

[8] D. J. Warren et al., “Recording and decoding for neural prostheses,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2507180. [9] P. Laguna Lasosa, J. P. Martı´nez, and E. Pueyo, “Techniques for ventricular repolarization instability assessment from the ECG,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015.2500501. [10] M. Baumert, P. Sanders, and A. N. Ganesan, “Quantitative-electrogram-based methods for guiding catheter ablation in atrial fibrillation,”

Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015.2505318. [11] K. Hosokawa and K. Sunagawa, “Closed-loop neuromodulation technology for baroreflex blood pressure control,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/ JPROC.2015.2496290. [12] A. E. W. Johnson et al., “Machine learning and decision support in critical care,” Proc. IEEE, vol. 104, no. 2, Feb. 2016, DOI: 10.1109/JPROC.2015.2501978.

ABOUT THE GUEST EDITORS Mathias Baumert (Senior Member, IEEE) received the Ph.D. degree from the Ilmenau University of Technology, Germany, in 2005. Since 2006 he has been with the University of Adelaide, Adelaide, S.A., Australia, holding appointments at the School of Electrical and Electronic Engineering, the School of Medicine and the School of Paediatrics and Reproductive Health. In 2014, he was appointed Associate Professor at the School of Electrical and Electronic Engineering. He is working on translating biomedical signal processing within the clinical setting, with focus on overnight polysomnography and electrocardiography. His interests are heart–brain interaction, sleep and arrhythmia. He has authored over 70 peer-reviewed articles in a wide range of international academic journals in the fields of biomedical engineering, cardiology, physiology, and neuroscience. Dr. Baumert won Australian Postdoctoral Fellowship (2006–2008) and Australian Research Fellowship (2011–2015) from the Australian Research Council as well as a Career Development Award (2011–2014) from the National Health & Medical Research Council Australia. According to Google Scholar his work has been cited 9 1700 times and his h-index is 23. Alberto Porta (Member, IEEE) received the M.S. degree in electronic engineering and the Ph.D. degree in biomedical engineering from the Politecnico of Milan, Milan, Italy, in 1989 and 1999, respectively. He was a Research Fellow on Automatic Control and System Theory at the University of Brescia, Brescia, Italy, until 1994. Since 1999, he has been with the Faculty of Medicine, University of Milan, Milan, Italy, where he became Researcher in 2005. Since 2011, he has been an Associate Professor in the same Faculty and University. Since 2006, he has been teaching medical physics, since 2007 applied medical statistics, and since 2011 bioengineering in the Faculty of Medicine, University of Milan. His primary interests include time-series analysis, spectral estimation, complexity, entropy, causality, prediction, nonlinear dynamics, biological cybernetics, system identification, and modeling. Applications are

222

Proceedings of the IEEE | Vol. 104, No. 2, February 2016

mainly focused on cardiovascular systems, cardiovascular regulatory mechanisms, and cardiovascular neuroscience. He is the author of more than 185 papers published in peer-review international journals. Dr. Porta is a reviewer of several international journals in the fields of computational biology, biomedical engineering, biological cybernetics, bioinformatics, applied physics and statistics, physiology, clinical cardiology, and neurosciences. Since 2007, he has been an Associate Editor of the International Conference of the IEEE Engineering in Medicine and Biology Society (Theme 1). He is currently a member of the board of the American Journal of Physiology (Regulatory, Integrative, and Comparative Physiology), Autonomic Neuroscience: Basic and Clinical, Clinical Autonomic Research, Frontiers in Autonomic Neuroscience, The European Physical Journal Nonlinear Biomedical Physics, and Physiological Measurement. He is the current President of the European Study Group on Cardiovascular Oscillations (ESGCO). His current h-index is 42 (source: Scopus). Andrzej Cichocki (Fellow, IEEE) received the M.Sc. (with honors), Ph.D., and Dr.Sc. (Habilitation) degrees in electrical engineering from Warsaw University of Technology, Warsaw, Poland. He spent several years at the University of Erlangen—Nuremberg, Erlangen, Germany, as an Alexander-von-Humboldt Research Fellow and Guest Professor. He is currently a Senior Team Leader and Head of the laboratory for Advanced Brain Signal Processing at RIKEN Brain Science Institute, Japan and full Professor at SKOLTECH. He is the (co)author of more than 400 technical journal papers and four monographs in English (two of them translated to Chinese). Currently, his research focuses on tensor decompositions, brain–computer interface, EEG hyperscanning, big data analytics, and their practical applications to rehabilitation and therapies. Dr. Cichocki was/is an Associated Editor of the IEEE TRANSACTIONS ON SIGNALS PROCESSING, the IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, the IEEE TRANSACTION ON CYBERNETICS, and the Journal of Neuroscience Methods, and was a founding Editor-in-Chief of the Journal Computational Intelligence and Neuroscience. His publications currently report over 26000 citations according to Google Scholar, with an h-index of 70.