Heart Disease Detection using Iridology with ANN

Abstract

Heart diseases are leading cause of death for men, and women around the world. Traditionally, to detect heart disease, heart condition checking is a must, but this method is costly, inconvenient, and takes some time. An alternative and a simpler method is the iridology method. Iridology is a study of the human iris to determine any abnormalities that happened in the organ’s functions. This study presents an implementation of computerized iridology in detecting heart disease. The system is designed with several stages such as pre-processing, segmentation region of interest, feature extraction, and classification using an Artificial Neural Network. Gray Level Co-Occurrence Matrix (GLCM) is used in feature extraction to extract the features from the segmented image while the Artificial Neural Network Backpropagation algorithm used as a classifier to create the prediction model for the system. The prediction model was evaluated using the 10-Fold Cross-Validation method. 50 patient data with 27 patients of a normal heart condition and another 23 patients of abnormal heart condition was used and the data been divided into 45 training data (90%) and 5 testing data (10%). The highest classification accuracy obtained is 95.56%.