Potential of Heart Disease Detection Based on Iridology

Every year the number of deaths caused by heart disease continues to increase. This is mostly caused by misdiagnosis or misinterpretation of the heart disease symptoms. Therefore, people need to be aware of this disease by maintaining a healthy lifestyle and conducting regular checks on the potential for heart disease. This early examination can be done using the iridology method, namely by analyzing the iris of the eye. This paper presents the implementation of computer vision and machine learning of iridology to detect the potential for heart disease. This system uses Canny edge detection and Principal Component Analysis (PCA) to extract features in the iris region of the eye, and Backpropagation Algorithm of Artificial Neural Networks to create the predictive model. There are 110 data used in this system, consisting of 55 eye images from subjects suffering from heart disease and 55 images of normal subjects. The data is divided into 88 data for training and 22 data for testing. The proposed system produced accuracy up to 95.45% for the test data using the sigma 0.3 for canny edge detection, 50 principal components, 50 hidden neurons, and 0.01 for the error limits.