A Parametric Framework for Modelling of Bioelectrical Signals

Extraction of useful information from cardiac signals for the diagnosis of diseasesand judgment of heart function is of special interest to medical personnel. Thus,the development of effective, robust, and efficient diagnostic tools for heart diseases isrequired. The aim when developing new techniques and tools is to minimize therequired cost and long hospitalization time, and increase the patient’s ease and safety. Inaccordance with this statement, in this PhD thesis, non-invasive electrical-basedmethods are of special interest. However, extracting useful information from measuredbiomedical data is not always trivial. The research community, including our previouscontributions, has developed many algorithms for separating the signals of differentorigins, e.g., cardiac, respiratory, and muscular activities, etc. Nevertheless, none of theexisting methods provides any mechanism to evaluate the performance of the developedalgorithms. Thus, there exist uncertainties regarding the properties of the signals, suchas its amplitude, waveform, components, and the origin of the signal waveform, which,in turn, limits the quality of the diagnostics of diseases and conditions.In this PhD thesis, it is argued that modelling the measured signals offers severaladvantages to help dealing with the above problems, as compared to relying onmeasured data only. By using a formalized representation, the parameters of the signalmodel can be easily manipulated and/or modified, thus providing mechanisms thatallow researchers to reproduce and control such signals.In turn, having such a formalized signal model makes it possible to developcomputer tools that can be used for manipulating and understanding how the signalchanges depend on various heart conditions, as well as for generating input signals forexperimenting with and evaluating the performance of, e.g. useful signal extractionmethods.In this work, the focus is on bioelectrical information, mainly electrical bioimpedance(EBI). Once the EBI is measured, it is necessary to model the correspondingsignals for analysis. In this case, the so-called advanced user should have to follow astructured approach to move from real measured data to the model of the correspondingsignals. For this, a generic framework is proposed in the PhD work. It has been used toguide the modelling of the impedance cardiography (ICG) and impedance respirography(IRG) signals. Here, based on statistical parameters and visual fit, a Fourier series isselected to model the ICG and IRG signals.The proposed framework has been used to guide the development of thecorresponding bio-impedance signal simulator (BISS). The internal details of thesimulator are presented, including the various model parameters and the mechanisms foradding modulation, noise, and motion artefacts. As a result, the implemented BISSgenerates simulated EBI signals and BISS gives freedom to the end-user to control theessential properties of the generated EBI signals depending on his/her needs. Predefinedhuman conditions/activities states are also included for ease of use.