Autonomous auscultation of the human heart

[摘要] ENGLISH ABSTRACT: The research presented in this thesis serves to provide a tool to autonomouslyscreen for cardiovascular disease in the rural areas of Africa. Vital informationthus obtained from patients can be communicated to advanced medical centres byTelemedicine. Cardiovascular disease is then detected in its initial stages, which isessential to its effective treatment. The system developed in this study uses recordedheart sounds and electrocardiogram signals to distinguish between normaland abnormal heart conditions. This system improves on standard diagnostic toolsin that it does not require cumbersome and expensive imaging equipment or ahighly trained operator.Heart sound- and electrocardiogram signals from 62 volunteers were recordedwith the prototype Precordialcardiogram device as part of a clinical study to aid inthe development of the autonomous auscultation software and to screen patientsfor cardiovascular disease. These volunteers consisted of 28 patients of TygerbergHospital with cardiovascular disease and, for control purposes, 34 persons withnormal heart conditions.The autonomous auscultation system developed during this study, interpretsdata obtained with the Precordialcardiogram device to autonomously acquire anormal or abnormal diagnosis. The system employs wavelet soft thresholding todenoise the recorded signals, followed by the segmentation of heart sound byidentifying peaks in the electrocardiogram. Novel frequency spectral informationwas extracted as features from the heart sounds, by means of ensemble empiricalmode decomposition and auto regressive modelling. These features proved to beparticularly significant and played a major role in the screening capability of thesystem. New time domain based features were identified, established on the specificcharacteristics of the various cardiovascular diseases encountered during thestudy. These features were extracted via the energy ratios between different partsof ventricular systole and diastole of each recorded cardiac cycle.The respective features were classified to characterise typical heart diseases aswell as healthy hearts with an ensemble artificial neural network. Herein the decisionsof all the members were combined to obtain a final diagnosis. The performanceof the autonomous auscultation system used in concert with the Precordialcardiogramdevice prototype, as determined through the leave-one-out crossvalidationmethod, had a sensitivity rating of 82% and a specificity rating of 88%.These results demonstrate the potential benefit of the Precordialcardiogram deviceand the developed autonomous auscultation software in a Telemedicine environment.

[发布机构] Stellenbosch University