Title: Machine learning approach to detect congenital heart diseases using palmar dermatoglyphics

Authors: Y. Mahesha; C. Nagaraju

Addresses: The National Institute of Engineering, Visvesvaraya Technological University, Mysore, Karnataka, India ' The National Institute of Engineering, Visvesvaraya Technological University, Mysore, Karnataka, India

Abstract: The present article has proposed a machine learning method to detect congenital heart diseases (CHDs) such as atrial septal defect (ASD) and myocardial infarction (MI) based on the frequency of occurrence of palm patterns such as ulnar loop and whorl. The system has been developed based on SSD-MobileNet to detect ulnar loop and whorl patterns on palm image. The developed system has achieved an accuracy of 99.28% and 97.19% in the detection of ulnar loop and whorl respectively. Further, the work has been carried out to fix the threshold value on the number of ulnar loop and whorl patterns to detect CHDs such as ASD and MI. The receiver operating characteristic curve has been drawn and the area under curve is calculated for the detection of ASD and MI. These results have shown that the proposed method can be used as a screening model to detect ASD and MI.

Keywords: ulnar loop; whorl; atrial septal defect; ASD; myocardial infarction; SSD-MobileNet.

DOI: 10.1504/IJMEI.2023.132575

International Journal of Medical Engineering and Informatics, 2023 Vol.15 No.4, pp.336 - 351

Received: 26 Mar 2021
Accepted: 05 Jun 2021
Published online: 30 Jul 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article