Cardiac seismocardiography analysis using 2- elements accelerometer sensor array and beamforming technique

• Main Author: Ng, Seng Hooi
• Format: Thesis
• Language: English
• Published: 2021
• Subjects: TJ Mechanical engineering and machinery
• Online Access: http://eprints.utm.my/id/eprint/101909/1/NgSengHooiMSKM2021.pdf; http://eprints.utm.my/id/eprint/101909/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:148434
• Institution: Universiti Teknologi Malaysia

Human heart contains a lot of informations that indicate the condition of its operation and health. The informations can be extracted using image, acoustic, electric and vibration signal. The problem with current technology is that it suffers badly with noise and other unwanted interference. To address this noise issue with the latest technology is echocardiography, a diagnostic tool for diagnosing on cardiac contractility and valvular disease. However, this device is quite costly and labour intensive which requires a specialist who is expert and enough experience in using this equipment. Furthermore, most of medical institutes unable to afford the cost of equipment facility. This study aimed to investigate the application of a non-invasive cardiac diagnostic approach using an accelerometer sensor array, coupled with a directional filtering approach to remove the unwanted noise. This work proposed the utilization of directional filtering method to remove noise using body vibration sensor by employing adaptive beamforming method without altering the signal information. Seismocardiography (SCG) was used to capture body vibration signals recorded via vibration sensor that collects information related to the heart pumping activities and later diagnosed the heart disease. The sensor array was used to collect SCG signal for 28 cycle data from normal and abnormal heart conditions of subjects in supine position. It was found that signal of heart disease information in SCG was overlapped with the noise signal. A directional denoising method which comprised of Delay and Sum (DAS) beamforming and Linearly Constrained Minimum Variance (LCMV) beamforming algorithm were applied, and the performance were compared. The result of signal to noise ratio (SNR) for DAS beamforming algorithm on normal subject was 7.11dB and abnormal subject was 4.13dB. For LCMV beamforming algorithm, normal subject was 10.85dB and abnormal subject is 7.04dB. Based on these results, it showed that the LCMV beamforming performed better than DAS as indicated in the SNR improvement by 30%. This SNR improvement represents the better accuracy of heart disease diagnosis.