Cholesterol Detection Through Iris Image Using Local Binary Pattern Method And Artificial Neural Network Classification


Cholesterol is fat that is in human blood which is needed for the formation of several hormones and new cell walls. Excess cholesterol in the blood can lead to coronary heart disease and stroke. Normal human cholesterol levels are in the range of 200 mg / dL or less. At this time, testing for cholesterol levels is still using blood or so-called invasive techniques. Previously, patients had to fast for 10-12 hours, making this technique less time-efficient. Therefore, a system was designed to detect cholesterol in a short time using eye images using iridological techniques. This study aims to detect a person’s cholesterol, including normal cholesterol, cholesterol risk, and high cholesterol. The system process begins with image data acquisition, then preprocessing is carried out which consists of resizing, ROI cropping and converting the RGB eye image to grayscale. In this study, the Local Binary Pattern (LBP) method is used as a feature extraction method and the Artificial Neural Network (ANN) method is used as its classification. Based on the test results, the system built is able to detect cholesterol through iris images and classify it into three classes. The number of iris images used was 120 images with 60 training data and 60 test data. The results obtained are the highest accuracy rate of 91.66% and an average computation time of 0.3362 s with the first order test parameters, radius (r) = 1, resize pixels 768 × 768, ROI = 64, epoch 1000 and hidden layer 10.

Authors: Agata Elisabet, Rita Magdalena, Jangkung Raharjo

Source: eProceedings of Engineering, Telcom University, Indonesia :