Monitoring and Advance Warning for Cardiac Arrhythmia Using PCG & ECG 6a.002.TUT


Project Start Date: Jul 1, 2017
Research Areas: Analytics, Analytics - Deep Learning, Analytics - Machine Learning, Analytics - Signal Processing, Data Management, Data Management - Big Data Platforms, Data Management - Cloud Computing, Data Management - Dynamic Data
Funding: Member Funded
Project Tags: ,

Project Summary

Auscultation of heart, besides the advanced heart diagnostic methods, continues to play an important role. The high correlation between phonocardiographic events and the characteristics of the hemodynamic affirm the heart sound crucial role in outpatient monitoring. In this project, we focus deliberately on the anomaly detection of heart sounds by using 1D Convolutional Neural Networks (CNN) trained with a novel data purification approach. The experimental results over the PhysioNet (CinC) Challenge 2016 benchmark dataset show that the proposed approach achieves high detection performance and a real-time processing ability under the condition that a reasonable signal quality (SNR) is present.

PERSONNEL

Principal Investigator(s)

Student(s)