Hybrid quantum-classical deep learning framework for balanced multiclass diabetic retinopathy classification

Sep 29, 2026

Upload by Stanley

Diabetic retinopathy (DR) is a progressive eye disease and a major cause of preventable blindness, making early and accurate severity-stage classification essential yet difficult due to class imbalance, high-resolution inputs, and limited scalability of existing models. A hybrid quantum-classical deep learning framework is proposed for five-class DR classification, targeting computational efficiency and class-balanced learning. The method reports 80.96% balanced accuracy on the APTOS 2019 dataset and is designed for telemedicine and low-resource clinical settings, supported by replicable code.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Abstract

Background

Disease motivation and global burden

Conventional diagnosis and ML/DL progress

Proposed framework (method overview)

MethodsX 15 (2025) 103605 Contents lists available at ScienceDirect MethodsX journal [homepage:](homepage: www.elsevier.com/locate/methodsx)[ www.elsevier.com/locate/methodsx](homepage: www.elsevier.com/locate/methodsx) | Hybrid quantum-classical deep learning framework for balanced multiclass diabetic retinopathy classification | | | --- | --- | | Tabassum Araa,b,* , Ved Prakash Mishra a, Manish Balia, Anuradha Yenkikara,c
a School of Engineering, Amity University Dubai Campus, Dubai, 25314, UAE
b Department of AI/ML, HKBK College of Engineering, Bengaluru, Karnataka, India c Department of CSE(AI), Vishwakarma Institute of Technology, Pune, 411048, Maharashtra, India | | | G R A P H I C A L A B S T R A C T | | | | | A R T I C L E I N F O Keywords: Diabetic retinopathy detection Quantum machine learning, Hybrid quantumclassical model ResNet50 Multiclass medical image classification A B S T R A C T Diabetic Retinopathy (DR) is a progressive eye disease and a leading cause of preventable blindness among diabetic patients. Early and accurate classification of its severity stages is crucial for effective treatment but remains challenging due to class imbalance, high-resolution data, and limited scalability of existing models. This study presents a novel hybrid quantum-classical deep learning framework to address these limitations in five-class DR classification. The model achieves a balanced accuracy of 80.96 % on the APTOS 2019 dataset, outperforming several classical baselines across all DR stages. It is optimized for computational efficiency and class-balanced learning, making it suitable for deployment in telemedicine platforms and low-resource clinical settings. This work contributes a scalable AI-based diagnostic approach that fuses deep learning with emerging quantum computing techniques. The methodology, results, and publicly shared codebase provide a replicable framework for researchers and practitioners working in AI for medical imaging and early disease screening. This method is well-suited for low-resource clinical environments and tele-ophthalmology applications. The method involves an: • ResNet-50 feature extractor with a 4-stage dense projection (2048→8) for quantum-ready * Corresponding author. E-mail address: [tabuara@gmail.com](tabuara@gmail.com) (T. Ara). [https://doi.org/10.1016/j.mex.2025.103605](https://doi.org/10.1016/j.mex.2025.103605) Received 28 June 2025; Accepted 2 September 2025 Available online 4 September 2025 2215-0161/© 2025 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ([http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/)). T. Ara et al. MethodsX 15 (2025) 103605 compression • 8-qubit VQC with parameterized RY–RZ gates and ring-style entanglement for high expressiveness • Stratified sampling + mixed-precision training for efficiency and class-balanced generalization | Related research article
None
Specifications table | | | --- | --- | | Subject area | Computer Science | | More specific subject area | Diabetic Retinopathy Detection | | Name of your method | Hybrid Quantum-classical DL model for balanced accuracy on five-class DR progression task | | Name and reference of original method | None | | Resource availability | PennyLane github, Dataset, Github | Background Diabetic Retinopathy (DR) is a progressive microvascular complication of diabetes mellitus that affects retinal blood vessels and may lead to irreversible vision loss if left undetected or untreated. The disease typically advances through distinct stages, from mild non-proliferative abnormalities to severe proliferative DR characterized by neovascularization eventually causing macular edema, hemorrhage, and retinal detachment [1], as depicted in Fig. 1(a). The global burden of DR is rising, with the International Diabetes Federation estimating that over 783 millio

FAQs

What problem does the framework address in diabetic retinopathy classification?

It targets the difficulty of early, accurate five-class DR severity classification under class imbalance, high-resolution data constraints, and limited scalability of existing models.

How does the proposed hybrid quantum-classical approach incorporate quantum learning?

It uses a ResNet-50 based feature extractor to produce quantum-ready features followed by an 8-qubit VQC with parameterized RY–RZ gates and ring-style entanglement.

Where is the method intended to be deployed and why?

The framework is optimized for computational efficiency and class-balanced generalization, making it suitable for deployment in telemedicine platforms and low-resource clinical environments.

Sep 29, 2026

Upload by Stanley