Verification of Iridology in Determining Dysfunctionality of Heart Through Deep-Learning

• Published in: International Conference on Signal Processing and Communication (Online) pp. 538 - 543
• Main Authors: Sruthi, K, Vijayakumar, J, Thavamani, S
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2022
• Subjects: classification; CNN; CVD; Deep learning; Heart; Image segmentation; iridology; Iris; Neural networks; Sensitivity
• ISSN: 2643-444X
• DOI: 10.1109/ICSC56524.2022.10009109

Cardio Vascular Disease (CVD) is a condition that occurs due to heart failure. CVD is one of the significant causes of life loss. If CVD left untreated, it may result in multiple organ failures. Iridology, a class of Complementary and Alternative Medicine, and a non-invasive diagnosing method, has the ability to determine the dysfunctional organ in human. In iridology dysfunctional organ is identified by analyzing the characteristics of iris. Deep learning is one of the promising methods in diagnosing various health-related issues. In this study, an attempt has been made to verify the efficiency of the iridology in determining subjects with heart disorder using the deep learning algorithm. Iris images from 133 subjects, 50 subjects having heart issues (unhealthy) and 83 subjects with normal functioning of heart (healthy), were obtained using Cogent CIS 202 iris scanner. The proposed system used Circular Hough transform and Daugman's rubber sheet model for iris segmentation and normalization, respectively. Based on iridology chart the region of interest, heart, were cropped out from normalized image. A seven-layered Convolution Neural Network (CNN) based deep learning model is formulated for categorizing subjects with unhealthy or healthy. The proposed system achieved accuracy, precision, specificity, and sensitivity of 95.839%, 92.905%, 97.289%, and 92.697%, respectively. Our study results verified the efficiency of iridology in identifying healthy and unhealthy subjects based on functionality of heart using deep learning.