Thesis (MScEng)--Stellenbosch University, 2014. / ENGLISH ABSTRACT: This thesis investigates the feasibility of the laser Doppler vibrometer (LDV) for use in the autonomous auscultation of the human heart. As a non-contact measurement device, the LDV could become a very versatile biomedical sensor. LDV, stethoscope, piezoelectric accelerometer (PA) and electrocardiogram (ECG) signals were simultaneously recorded from 20 volunteers at Tygerberg Hospital. Of the 20 volunteers, 17 were confirmed to have cardiovascular disease. 3 patients with normal heart sounds were recorded for control data. The recorded data was successfully denoised using soft threshold wavelet denoising and ensemble empirical mode decomposition. The LDV was compared to the PA in common biomedical applications and found to be equally accurate. The heart sound cycles for each participant were segmented using a combination of ECG data and a simplicity curve. Frequency domain features were extracted from each heart cycle and input into a k-nearest neighbours classifier. It was concluded that the LDV can form part of an autonomous, non-contact auscultation system. / AFRIKAANSE OPSOMMING: Hierdie tesis ondersoek die haalbaarheid daarvan om die laser Doppler vibrasiemeter (LDV) vir die outonome beluistering van die menslike hart te gebruik. As 'n kontaklose meettoestel kan die LDV werklik 'n veelsydige biomediese sensor word. Twintig vrywilligers by die Tygerberg Hospitaal se LDV-, stetoskoop-, piësoelektriese versnellingsmeter (PV)- en elektrokardiogram (EKG) seine is gelyktydig opgeneem. Uit die 20 vrywilligers was daar 17 bevestigde gevalle van kardiovaskulêre siektes. Die data van drie pasiënte met normale hartklanke is as kontroledata opgeneem. Geraas is suksesvol uit die opgeneemde data verwyder deur 'n kombinasie van sagtedrempelgolf en saamgestelde empiriese modus ontladingstegnieke. Die LDV was vergelyk met die PV vir algemene biomediese gebruike en daar was gevind dat dit vergelykbare akkuraatheid het. Die hartklanksiklusse van elke deelnemer is gesegmenteer deur EKG data en 'n eenvoudskromme te kombineer. Frekwensiegebiedskenmerke is uit elke hartsiklus onttrek en in 'n k-naastebuurpunt klassifiseerder ingevoer. Daar is tot die gevolgtrekking gekom dat die LDV deel van 'n outonome, kontaklose beluisteringstelsel kan uitmaak.

Cette thèse dans le domaine du traitement des signaux non-stationnaires, appliqué aux bruits du cœur mesurés avec un stéthoscope numérique, vise à concevoir un outil automatisé et « intelligent », permettant aux médecins de disposer d’une source d’information supplémentaire à celle du stéthoscope traditionnel. Une première étape dans l’analyse des signaux du cœur, consiste à localiser le premier et le deuxième son cardiaque (S1 et S2) afin de le segmenter en quatre parties : S1, systole, S2 et diastole. Plusieurs méthodes de localisation des sons cardiaques existent déjà dans la littérature. Une étude comparative entre les méthodes les plus pertinentes est réalisée et deux nouvelles méthodes basées sur la transformation temps-fréquence de Stockwell sont proposées. La première méthode, nommée SRBF, utilise des descripteurs issus du domaine temps-fréquence comme vecteur d’entré au réseau de neurones RBF qui génère l’enveloppe d’amplitude du signal cardiaque, la deuxième méthode, nommée SSE, calcule l’énergie de Shannon du spectre local obtenu par la transformée en S. Ensuite, une phase de détection des extrémités (onset, ending) est nécessaire. Une méthode d’extraction des signaux S1 et S2, basée sur la transformée en S optimisée, est discutée et comparée avec les différentes approches qui existent dans la littérature. Concernant la classification des signaux cardiaques, les méthodes décrites dans la littérature pour classifier S1 et S2, se basent sur des critères temporels (durée de systole et diastole) qui ne seront plus valables dans plusieurs cas pathologiques comme par exemple la tachycardie sévère. Un nouveau descripteur issu du domaine temps-fréquence est évalué et validé pour discriminer S1 de S2. Ensuite, une nouvelle méthode de génération des attributs, basée sur la décomposition modale empirique (EMD) est proposée.Des descripteurs non-linéaires sont également testés, dans le but de classifier des sons cardiaques normaux et sons pathologiques en présence des souffles systoliques. Des outils de traitement et de reconnaissance des signaux non-stationnaires basés sur des caractéristiques morphologique, temps-fréquences et non linéaire du signal, ont été explorés au cours de ce projet de thèse afin de proposer un module d’aide au diagnostic, qui ne nécessite pas d’information à priori sur le sujet traité, robuste vis à vis du bruit et applicable dans des conditions cliniques. / This thesis in the field of biomedical signal processing, applied to the heart sounds, aims to develop an automated and intelligent module, allowing medical doctors to have an additional source of information than the traditional stethoscope. A first step in the analysis of heart sounds is the segmentation process. The heart sounds segmentation process segments the PCG (PhonoCardioGram) signal into four parts: S1 (first heart sound), systole, S2 (second heart sound) and diastole. It can be considered one of the most important phases in the auto-analysis of PCG signals. The proposed segmentation module in this thesis can be divided into three main blocks: localization of heart sounds, boundaries detection of the localized heart sounds and classification block to distinguish between S1and S2. Several methods of heart sound localization exist in the literature. A comparative study between the most relevant methods is performed and two new localization methods of heart sounds are proposed in this study. Both of them are based on the S-transform, the first method uses Radial Basis Functions (RBF) neural network to extract the envelope of the heart sound signal after a feature extraction process that operates on the S-matrix. The second method named SSE calculates the Shannon Energy of the local spectrum calculated by the S-transform for each sample of the heart sound signal. The second block contains a novel approach for the boundaries detection of S1 and S2 (onset & ending). The energy concentrations of the S-transform of localized sounds are optimized by using a window width optimization algorithm. Then the SSE envelope is recalculated and a local adaptive threshold is applied to refine the estimated boundaries. For the classification block, most of the existing methods in the literature use the systole and diastole duration (systole regularity) as a criterion to discriminate between S1 and S2. These methods do not perform well for all types of heart sounds, especially in the presence of high heart rate or in the presence of arrhythmic pathologies. To deal with this problem, two feature extraction methods based on Singular Value Decomposition (SVD) technique are examined. The first method uses the S-Transform and the second method uses the Intrinsic Mode Functions (IMF) calculated by the Empirical Mode Decomposition (EMD) technique. The features are applied to a KNN classifier to estimate the performance of each feature extraction method. Nonlinear features are also tested in order to classify the normal and pathological heart sounds in the presence of systolic murmurs. Processing and recognition signal processing tools based on morphological, time-frequency and nonlinear signal features, were explored in this thesis in order to propose an auto-diagnosis module, robust against noise and applicable in clinical conditions.

For this project, the design, implementation, characterization, calibration and possible applications of a multiple transducer stethoscope apparatus were investigated. The multi-transducer sensor array design consists of five standard stethoscope diaphragms mounted to a rigid frame for a-priori knowledge of their relative spatial locations in the x-y plane, with compliant z-direction positioning to ensure good contact and pressure against the subject’s skin for reliable acoustic coupling. When this apparatus is properly placed on the body, it can digitally capture the same important body sounds investigated with standard acoustic stethoscopes; especially heart sounds. Acoustic signal inputs from each diaphragm are converted to electrical signals through microphone pickups installed in the stethoscope connective tubing; and are subsequently sampled and digitized for analysis. With this system, we are able to simultaneously interrogate internal body sounds at a sampling rate of 2 KHz, as most heart sounds of interest occur below 200 Hz. This system was characterized and calibrated by chirp and impulse signal tests. After calibrating the system, a variety of methods for combining the individual sensor channel data to improve the detectability of different signals of interest were explored using variable-delay beam forming. S1 and S2 heart sound recognition with optimized beam forming delays and inter-symbol noise elimination were investigated for improved discernment of the S1 or S2 heart sounds by a user. Also, stereophonic presentation of heart sounds was also produced to allow future investigation of its potential clinical diagnostic efficacy.

This thesis deals with the possibility of using impedance cardiography for calculating cardiac output. Kubicek’s, Sramek‘s and Sramek-Bernstein‘s methods are discussed here. These methods were applied to a data set, obtained by measuring on subjects with implanted cardiostimulators. The subjects’ heart rate was being changed by the programing of cardiostimulators. Thanks to this procedure the measured data were not affected by artifacts, connected with the heart rate change caused by a body stress, or other influences. An influence of heart rate on a cardiac output value based on the statistical processing of the data set was studied.

In developing countries, acute respiratory infections (ARIs) are responsible for two million deaths per year. Most victims are children who are less than 5 years old. Pneumonia kills 5000 children per day. The statistics for cardiovascular diseases (CVDs) are even more alarming. According to a 2009 report from the World Health Organization (WHO), CVDs kill 17 million people per year. In many resource-poor parts of the world such as India and China, many people are unable to access cardiologists, pulmonologists, and other specialists. Hence, low skilled health professionals are responsible for screening people for ARIs and CVDs in these areas. For example, in the rural areas of the Philippines, there is only one doctor for every 10,000 people. By contrast, the United States has one doctor for every 500 Americans. Due to advances in technology, it is now possible to use a smartphone for audio recording, signal processing, and machine learning. In my thesis, I have developed an Android application named Smart Auscultation. Auscultation is a process in which physicians listen to heart and lung sounds to diagnose disorders. Cardiologists spend years mastering this skill. The Smart Auscultation application is capable of recording and classifying heart sounds, and can be used by public or clinical health workers. This application can detect abnormal heart sounds with up to 92-98% accuracy. In addition, the application can record, but not yet classify, lung sounds. This application will be able to help save thousands of lives by allowing anyone to identify abnormal heart and lung sounds.

No description available.