Feature Extraction For Application of Heart Abnormalities Detection Through Iris Based on Mobile Devices


As the WHO says, heart disease is the leading cause of death and examining it by current methods in hospitals is not cheap. Iridology is one of the most popular alternative ways to detect the condition of organs. Iridology is the science that enables a health practitioner or non-expert to study signs in the iris that are capable of showing abnormalities in the body, including basic genetics, toxin deposition, circulation of dams, and other weaknesses. Research on computer iridology has been done before. One is about the computer’s iridology system to detect heart conditions. There are several stages such as capture eye base on target, pre-processing, cropping, segmentation, feature extraction and classification using Thresholding algorithms. In this study, feature extraction process performed using binarization method by transforming the image into black and white. In this process we compare the two approaches of binarization method, binarization based on grayscale images and binarization based on proximity. The system we proposed was tested at Mugi Barokah Clinic Surabaya. We conclude that the image grayscale approach performs better classification than using proximity.

The heart is our organ that controls the blood circulation. The heart disease leading to death. It can affect everyone, both young and old. Iridology  is an alternative medicine technique to detect heart abnormalities. Needs a process in feature extraction in an iris image to perform iridology and computation process. The mobile device is one of the applications platforms which is very popular right now. It could do a lot of computing that used to be only on the computer. So we can perform every computation in anywhere and anytime. This research is proposed a new method of iridology and computation process in order to check the heart condition. By applying the mobile devices, everyone can check their heart regularly.
The feature extraction method produces a high result of right classification while it performs on success image cropping. From that
experiment, the black threshold used is 0.725 and the white threshold used is0.275.
Febriana Dyah Kusuma ; Entin Martiana Kusumaningtyas ; Ali Ridho Barakbah ; Aditya Afgan Hermawan , Electronics Engineering Polytechnic Institute of Surabaya