Screening for target heart disease using discrete wavelet transform and artificial neural network techniques


Gokongwei College of Engineering

Document Type

Archival Material/Manuscript

Publication Date



The objective of this research is to show the diagnostic value of discrete wavelet transform (DWT) and artificial neural network (ANN) techniques in the analysis of cardiogram (ECG) signals. The results obtained may be used to develop equipment that can screen patients for target heart diseases. Easy access to such equipment is expected to reduce mortality rates of patients due to heart-related diseases.

The computer-based system was developed to interpret ECG signal using DWT and ANN techniques. Hardcopies of ECG records were scanned and converted to signals which were then processed using DWT techniques to extract feature parameters in the form of wavelet coefficients. The standard deviation of the wavelet coefficients was fed to the input of an ANN previously trained on four heart conditions. The heart condition diagnoses of the ECG traces were identified or confirmed through four output nodes of the ANN.

The system was able to confirm two heart conditions with a 70% success rate but failed to confirm a normal heart condition with a 20% success rate. Two major problems that were encountered in the development of the system were a scarcity of usable ECG records that prevented a larger training set for the ANN and an observed subjective interpretation by cardiologists.



Electrical and Computer Engineering


Wavelets (Mathematics); Neural networks (Computer science); Electrocardiography

Upload File


This document is currently not available here.