Andrei Krivoshei

An Adaptive Filtering System for Separation of Cardiac and Respiratory Components of Bioimpedance Signal

This paper presents an adaptive filtering system for separation of two bio-impedance signal components: cardiac and respiratory signals. The proposed filtering system is adaptive to the parameters of the input signals cardiac component (the reference signal), which is corrupted by the respiratory component and also by additive stochastic disturbances. The adaptation is achieved applying estimation and continuous tracking of the heart rate using a time-optimal adaptive phase-locked loop (APLL). Technical solutions of the filtering system are oriented on applications in portable and implantable medical devices

Decomposition method of an electrical bio-impedance signal into cardiac and respiratory components

The paper presents a method for adaptive decomposition of an electrical bio-impedance (BI) signal into two components: cardiac and respiratory. The decomposition of a BI signal is not a trivial process because of the non-stationarity of the signal components and overlapping of their harmonic spectra. An application specific orthonormal basis (ASOB) was designed to solve the decomposition task using the Jacobi weighting function in the standard Gram-Schmidt process. The key element of the bio-impedance signal decomposer (BISD) is a model of the cardiac BI signal, which is constructed from the components of the ASOB and is intended for use in the BISD for on-line tracking of the cardiac BI signal. It makes it possible to separate the cardiac and respiratory components of the total BI signal in non-stationary conditions. In combination with the signal-shape locked loop (SSLL), the BISD allows us to decompose the BI signals with partially overlapping spectra. The proposed BISD based method is accomplished as a PC software digital system, but it is oriented towards applications in portable and stationary cardiac devices and in clinical settings.

Non-invasive method for the aortic blood pressure waveform estimation using the measured radial EBI

The paper presents a method for the Central Aortic Pressure (CAP) waveform estimation from the measured radial Electrical Bio-Impedance (EBI). The method proposed here is a non-invasive and health-safe approach to estimate the cardiovascular system parameters, such as the Augmentation Index (AI). Reconstruction of the CAP curve from the EBI data is provided by spectral domain transfer functions (TF), found on the bases of data analysis. Clinical experiments were carried out on 30 patients in the Center of Cardiology of East-Tallinn Central Hospital during coronary angiography on patients in age of 43 to 80 years. The quality and reliability of the method was tested by comparing the evaluated augmentation indices obtained from the invasively measured CAP data and from the reconstructed curve. The correlation coefficient r = 0.89 was calculated in the range of AICAP values from 5 to 28. Comparing to the traditional tonometry based method, the developed one is more convenient to use and it allows long-term monitoring of the AI, what is not possible with tonometry probes.

Electrical Bio-Impedance based non-invasive method for the Central Aortic blood pressure waveform estimation

Non-invasive and safe method for estimation of the Central Aortic Pressure (CAP) waveform is presented. It is based on the measurement of Electrical Bio-Impedance (EBI) on radial artery. Reconstruction of the CAP curve from the EBI data is provided by an introduced spectral domain transfer function (TF). The TF estimation algorithm and the Augmentation Index (AI) evaluation method are described. Proposed approach is less dependent on the operator and more convenient than traditional tonometry based method. It allows continuous and long term monitoring of the AI and CAP variations. Feasibility and accuracy of the technique was evaluated in clinical surroundings by comparing invasively measured CAP and AI data with non-invasive EBI based estimates. Clinical experiments were conducted in the Heart Centre of the East-Tallinn Central Hospital during scheduled coronary angiography. Age of the patients varied randomly between 43 and 80.

Signal-Shape Locked Loop (SSLL) as an Adaptive Separator of Cardiac and Respiratory Components of Bio-Impedance Signal

The paper presents an on-line signal processing system for adaptive separation of two infra-low frequency signals: cardiac and respiratory bio-impedance (BI) signals, which are the time varying components of the total BI signal. The separation process of such signals as cardiac and respiratory BI components, is not a trivial filtering due to overlapping of spectra and non stationarity of these signals, and moreover, due to the infra-low frequency range. Therefore, advanced signal processing concepts and methods are needed to achieve the goal. The Signal-Shape Locked Loop (SSLL) concept was introduced to solve the task. Using this concept, it is possible to separate two (or more) independent signal components from the total input signal. Technical solution of the system is intended for applications in portable and implantable cardiac devices.

A bio-impedance signal simulator (BISS) for research and training purposes

A Bio-Impedance Signal Simulator (BISS) is developed based on the models of the impedance cardiography (ICG) and impedance respirography (IRG) signals. With the aim of imitating the real ICG and IRG phenomena, the ICG and IRG signals are modelled and combined with motion artefacts and Gaussian noise. The simulator allows the user to load different predefined human activity states such as resting, standing, walking, and running. Moreover, and importantly, the user can also control the parameters as per his/her needs and generate Electrical Bio-Impedance (EBI) datasets for further processing. Possible applications of BISS include research (e.g. performance evaluation of cardiac and respiratory separation algorithms) as well as teaching and training in physiological courses. To the best of our knowledge, BISS is the first EBI signal simulator that imitates the real ICG and IRG signals phenomena.

Excitation Current Source for Bioimpedance Measurement Applications: Analysis and Design

This paper presents a systematic current source analysis using the transfer function approach. The analysis is done for the so called load-in-loop current source circuit configuration, and is generalized also for the grounded load impedance. Presence of serial and parallel parasitic components is taken into account. The analytical results are used to define criteria for the current source design and to evaluate the expressions for the source parameters estimation. Moreover the load-in-loop current source circuit is proposed in the paper, together with the current source parameters measurement system configuration. In the load-in-loop current source circuit implementation, an inverting amplifier is used with the gain G= -1. Such a choice gives theoretically only two times bigger effective intrinsic impedance value, but is usable for comparing the measured data with the theoretical results and the results of simulation. For this case good accordance of experimental results with the proposed theoretical analysis and the simulation results was achieved.

An Adaptively Tunable Model of the Cardiac Signal for the Bio-Impedance Signal Decomposer (BISD)

The paper presents the further development of the bio- impedance signal decomposer (BISD) of the total bio-impedance (BI) signal to its cardiac and respiratory components. The Jacobi orthonormal polynomials based adaptively tunable model of the cardiac BI signal is proposed in the paper, which plays very important role in the decomposition task. The importance arises from the fact, that the BISD must be reliable and have to correct operate with signals taken from different persons, and in such cases, when the cardiac BI signal of a person is changing in time. For the proposed system the reliability significantly depends on the difference between the model of the cardiac signal and the real cardiac signal (the reference signal). The averaged through several periods version of the already separated cardiac BI signal is used as reference signal in the proposed algorithm for tuning the parameters of the cardiac BI signal model using a modified Newton adaptation algorithm. After the model is elaborated, the system separates the cardiac and the respiratory components more accurately by tracking the cardiac BI signal.

Bio-Impedance Signal Decomposer (BISD) as an Adaptive Signal Model Based Separator of Cardiac and Respiratory Components

The paper presents the Bio-Impedance Signal Decomposer (BISD) for adaptive separation of two infra-low frequency signals: cardiac and respiratory bio-impedance (BI) signals, which are the time varying components of the total electrical BI signal. The adaptively tuneable cardiac BI signal model (CBISM), constructed from components of a specially designed orthonormal basis is the key element of the proposed BISD. If combined with the signal-shape locked loop (SSLL) concept, this allows an on-line decomposition of a BI signal into its cardiac and respiratory components with partially overlapping spectra. The proposed BISD is currently designed and realised as a digital (PC software) system. The solution is also oriented to be used for portable and implantable cardiac devices, such as rate adaptive pacemakers as well as for stationary devices in clinical conditions.

CAP waveform estimation from the measured electrical bioimpedance values: Patient's heart rate variability analysis.

The paper presents analysis of the generic transfer function (TF) between Electrical Bioimpedance (EBI) measured non-invasively on the wrist and Central Aortic Pressure (CAP) invasively measured at the aortic root. Influence of the Heart Rate (HR) variations on the generic TF and on reconstructed CAP waveforms is investigated. The HR variation analysis is provided on a single patient data to exclude inter-patient influences at the current research stage. A new approach for the generic TF estimating from a data ensemble is presented as well. Moreover, an influence of the cardiac period beginning point selection is analyzed and empirically optimal solution for its selection is proposed.