Leveraging Quantum-Based Architectures for Robust Diagnostics

TL;DR

This paper introduces a hybrid classical-quantum diagnostic framework using quantum convolutional neural networks for multi-class medical image classification, achieving high accuracy and outperforming classical models.

Contribution

It presents a novel hybrid quantum-classical approach with dataset-specific preprocessing and transfer learning, demonstrating superior performance in medical image diagnostics.

Findings

Achieved 99% accuracy on kidney CT classification.

Outperformed classical CNN baselines in precision, recall, and F1 scores.

Demonstrated stable convergence across multiple medical imaging tasks.

Abstract

Quantum machine learning has emerged as a promising approach for medical image analysis, particularly in settings where compact models and expressive feature representations are desired. This paper presents a hybrid classical--quantum diagnostic framework that integrates dataset-specific preprocessing, transfer learning, and quantum convolutional neural networks (QCNNs) for multi-class medical image classification. This approach is evaluated on three distinct tasks: kidney disease diagnosis from computed tomography images, cervical cell classification from pap smear images, and brain tumor classification from magnetic resonance imaging. For each dataset, a pretrained encoder is used to extract latent features, which are then embedded into quantum states through angle or amplitude encoding and processed by a QCNN. Experimental results show strong and stable convergence across all datasets. The proposed hybrid models achieve 99% test accuracy on kidney CT classification, 97% on cervical cell classification, and 99% on brain tumor classification. In comparative evaluations for precision, recall, and F1, the hybrid QCNN models consistently outperform classical CNN baselines using the same pretrained encoders and similar hyperparameter settings, while requiring fewer trainable parameters. These results demonstrate the potential of quantum-enhanced architectures for robust and efficient medical diagnostics.