Evaluating the Impact of the Caputo Derivative on Deep Models for Hypertrophic Cardiomyopathy Classification Using CMR Images

Bushra K.  Hilal; Ebtesam N.  AlShemmary

doi:10.24996/ijs.2026.67.2.%g

Authors

Bushra K. Hilal Department of Computer Science College of Computer Science and Mathematics, University of Kufa, Kufa, Iraq / Department of Computer Information Systems, College of Computer Science and Information Technology, University Al- Qadisiyah, Al-Qadisiyah, Iraq https://orcid.org/0009-0002-7635-1141
Ebtesam N. AlShemmary IT Research and Development Center, University of Kufa, Kufa, Iraq https://orcid.org/0000-0001-7500-9702

DOI:

https://doi.org/10.24996/ijs.2026.67.2.%25g

Keywords:

Cardiac magnetic resonance (CMR), Caputo derivative, Deep learning models, EfficientNetV2S, Hypertrophic cardiomyopathy (HCM)

Abstract

To enhance patient outcomes, it is essential to diagnose hypertrophic cardiomyopathy (HCM) from cardiac magnetic resonance (CMR) images with precision, ensuring the process is swift and automated. This study investigates the impact of integrating Caputo derivatives into deep learning models to enhance their performance in classifying HCM. The study examines the performance of a tailored convolutional neural network (CNN), the advanced EfficientNetV2S architecture, and the improved CNN incorporating the Caputo derivative. Key pre-processing techniques included image resizing, normalization, and data augmentation. Caputo’s CNN performed best with 92.47% accuracy, 93.57% precision, and 89.36% F1 score with a slightly reduced recall of 85.68%, while EfficientNetV2S achieved the highest accuracy (98.62%), demonstrating exceptional feature extraction capabilities. The results suggest that fractional calculus combined with deep learning can deepen diagnostic accuracy in CMR while providing more effective and interpretable HCM classification frameworks.