Designing Diabetes Mellitus Detection System Based on Iridology with Convolutional Neural Network Modeling

Diabetes mellitus is one of the uncontagious diseases with the highest mortality rate in the world. It happens because of the increased risk of complications caused by the disease. One of the preventative ways is to do early detection, one of which is by using the iridology method. The method detects damage to the body’s organs through the signs that appear on the iris. The paper has introduced a Diabetes Mellitus Detection System to classify diabetes using a Convolutional Neural Network (CNN). The proposed method removed the pupil segmentation step that is important in the traditional machine learning classification system. The squared pupil image size 720×360 pixel was trained using Adam’s algorithm with a learning rate of 0.001 to develop the CNN model. The pupil image was collected using Iriscope Iris Analyzer Iridology 9822U camera. The dataset consists of 35 healthy and 14 diabetes subjects that repeat three times of each person. The proposed approach has an accuracy of 96.43% that better performance compared to traditional machine learning.

Authors: Dyla Velia; Adhi Harmoko Saputro