Abstract:

Coronary Artery Disease (CAD) is a heart disease caused by the narrowing or blockage of the coronary arteries that supply the heart muscle. Early diagnosis of CAD, a health problem with a high mortality rate worldwide, is very important. In this study, we aimed to predict CAD using iridology and image processing techniques. Unlike previous studies, the performance of the Local Binary Patterns (LBP) feature extraction method, which has not been utilized in iridology-based heart disease prediction studies, was analyzed. In the proposed method, features were extracted with LBP from iris images of a total of 198 volunteers (94 CAD and 104 Control group), and classification was performed using Artificial Neural Network (ANN). The Integral Differential Operator method was used to find the iris positions in the image, and Rubber Sheet Normalization was used to convert the iris into a rectangular format. Through the iridology map, the heart region in the iris was determined as the analysis region, and 59 histogram-based features were extracted from this region with one pixel and eight neighborhoods with the LBP. The classification was performed using ANN with the extracted features. The data were divided into two groups: training and test. The Scaled Conjugate Gradient (SCG) algorithm performed the training process. The accuracy, precision, sensitivity, specificity, F1 score and Area Under the Curve (AUC) values determined as performance criteria were 91.5%, 0.9063, 0.9355, 0.8929, 0.92063 and 0.9103 for the test data, respectively. Based on the findings, it can be said that the proposed method based on the LBP is successful in predicting CAD.

Keywords: