Hybrid Quantum-Classical Deep Learning for Chest X-ray Classification Using ResNet, MobileNet, and Quantum Models under Data-Constrained Conditions
Hybrid Quantum-Classical Deep Learning for Chest X-ray Classification Using ResNet, MobileNet, and Quantum Models under Data-Constrained Conditions
Authors:
Lankapalli Mahitha Chopra
MSc Artificial Intelligence and Data Science
Central University of Andhra Pradesh
Dr. C. Krishna Priya
Assistant Professor, Department of Computer Science and AI
Central University of Andhra Pradesh
Corresponding Author:
krishnapriyarams@cuap.edu.in
Abstract
Data scarcity and the computationally expensive nature of deep learning hinder its applications for medical image classification. This paper explored hybrid quantum-classical models for classifying chest X-rays in data-restricted environments. ResNet18, MobileNetV2, and its hybrid quantum variants were trained at 100%, 50%, 20% and 5% training sets respectively. The model was assessed by accuracy, F1-score, and the area under the curve (AUC). It was found that classical models had better accuracy compared to hybrid models, whereas hybrid models improved the AUC and hence class separability. In a low-data regime, MobileNet performed better due to its computationally lightweight nature, and the best performing Hybrid model combining efficiency with robustness is the MobileNet + Quantum model, revealing promising medical applications for the models where data is scarce.
Keywords: Quantum Machine Learning, Deep Learning, MobileNet, ResNet, Chest X-ray, Medical Imaging