ANALYSIS OF ELECTROCARDIOGRAM AND CARDIO-VASCULAR AND PULMONARY SYSTEMS

Abstract

This thesis deals with the work done in the areas of analysis of electrocardiogram and cardiovascular and pulmonary systems. Electrocardiogram is extensively used by cardiologists in clinical interpretations for knowing the functioning of the heart. Electrocardiogram has good correlation with the mechanical activity of the heart which is responsible for the flow of blood throughout the circulatory system. The pulmonary system is responsible for maintaining the proper level of Oxygen content inside the blood. Proper functioning of the cardiovascular system is dependent on the state of the cardiovascular and pulmonary systems. The work done for this thesis can be divided into two major sections° (i) Analysis of electrocardiogram (ii) Analysis of cardiovascular and pulmonary systems. New methods have been developed for the analysis of electrocardio gram. Existing models with modifications have been used for the analysis of cardiovascular and pulmonary systems. ANALYSIS OF ELECTRO CARDIOGRAM The first section deals with the accurate representation of ECG signal, derivation of diagnostically significant parameters, and development of the methodology for classification of cases according to nature of abnormality in ECG patterns. Twelve lead -iii- ECG system has been used for this work. Measurement error and lead proximity corrections are applied to lead potentials using Burger triangle representation. Peak potentials at leads I,II and III have been used to find the Frontal plane Peak Resultant vectors (FPR) for various segments of ECG pattern. The precisions achieved for amplitude and phase are 0.0000002 mV and 0.00O00Hdegree, respectively. From the set of normal cases, the different ranges of variation for various parameters are established. The relative valaes of parameters showed very good clustering in the normal pattern space. Sine, cosine, unit impulse, exponential and Gaussian functions have some limitations as the basis functions for the model representation of ECG waves. Therefore, a set of new mathe matically defined basis functions is proposed for various segments of ECG wave. These basis functions have specified shapes. The coefficients of the polynomial expressions defining the various basis functions are computed and stored once for all. The main variations in the ECG patterns are horizontal and vertical elong ation and/or contraction of various segments. There are also relative phase shifts among the segments and also lead to lead variations. All these have been properly taken into account in this model. In addition to the permanently stored model parameters the number of parameters required to generate the corrected pattern at the six frontal plane leads are less than or equal to eighteen. The set of ECG patterns has been defined by a mathematical expres sion. The ECG pattern at any frontal plane lead with measurement and lead proximity corrections can be generated by this model -ivexpression. There is a considerable data compression resulting in less memory storage requirement in computers. Simple and composite binary codes have been developed for representing the diagnostic parameters as symptoms. The parameters within and beyond the normal range are represented by a 0 and 1, respectively. For composite binary code, the parameters are represented as 11, for values within normal range, 01, for values below normal range and 10, for values above normal range. Symptom patterns are specified for four categories namely, normal case, myocardial infarction, .right ventricular hypertrophy and left ventricular hypertrophy. Depending upon the matching or mismatch ing of a symptom of a case with the specified symptom, the multi plier may be taken as 1, 0 or -1 for the respective symptom. Linear programming has been used to find the weight factors of symptoms with respect to different categories. The cases are classified into their respective categories by computing and comparing weighted sums of symptoms for a case with respect to different categories. The weighted sum is expressed as, N Y(I) = ( S W(J) * MM(I,J) 3=1 where, j corresponds to the symptom number, MM the symptom multi plier, I the category, and Wthe weight factor. The development of the methodology and computation of coefficients to be stored once for all has been carried out on DEC-20 computer. The computation for the final interpretation became quite simple and showed a possibility of microprocessor -Vimplementation. The software implementation of the method on a 8085 based microprocessor system is also given in this work. The method developed here using only three of the twelve leads is simple and efficient for ECG interpretation. The sample cases of myocardial infarction, left ventricular hypertrophy, right ventricular hypertrophy, and normal categories have been properly discriminated by this method. The microprocessor imple mentation shows that the method is well suited for bedside online analysis in hospitals. ANALYSIS OF CARDIOVASCULAR AND PULMONARY SYSTEMS The invasive measurements in the cardiovascular system are to be limited in number due to discomfort to the patient. This restricts the number of measurable physiological parameters. Therefore, with limited number of measured quantities a large number of diagnostically significant additional parameters for the system are found out using parameter estimation techniques. Clark et al.(1980) presented a modified Wind Kessel model of left heart systemic circulation system and proposed a method for its parameter estimation. This model has been used with modification in the present work for studying the effect of variation of aortic valve resistance, resistances and compliances of proximal and distal parts of left heart systemic circuit, peripheral resistance, inertance of long fluid columns and aortic valve switching func tions for systolic and diastolic periods. Maxima, minima and average values of responses are computed and considered for comparison purposes. The work has been done to establish -VIcorrelation among abnormalities of cardiovascular system, affected parameters and system responses. Simulation of certain abnormal states and associated mechanism is also possible. During the analysis a new source of mechanical arrhythmia has been observed. The means to arrest this arrhythmia are also discussed in this work. Large number of lumped parameter models of respiratory system has been developed so far. The respiratory system has been modelled using R-C or RLC parameters. In the present work RC and RLC models have been used for the analysis. The parameters have been estimated by the solution of the system equations for these models.This model analysis reduces the invasive measurements considerably. The loss of diagnostic information is compensated by the estimation of additional parameters. The results of the analysis are useful for knowing the state of overall pulmonary system and comparison of cases of different categories.