Cholesterol Level Detection through Eye Image Using Fractal and Decision Tree

Abstract

Cholesterol is a natural substance with physical properties of fat but has a steroid group. High cholesterol levels will cause hypertension, coronary heart disease. Cholesterol levels can now be detected through eye images. This research produces a cholesterol level detection system with input in the form of iris images. First, the image is resized, converted to grayscale, and cropped by the system. Then feature extraction is done by the fractal method, which has characteristics that can explain dimensions in non-integers. The last stage is a classification using the decision tree method because it can simplify a complex decision-making process to be more specific. Eye images are classified into three, namely cholesterol, cholesterol risk, and no cholesterol. In total, there are 105 images, consisting of 63 training data images and 42 test data images. The result is 95.23% accuracy, 90.47% precision, 100% recall, and 40 ms computing time.