Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Lightweight Medical Image Classification Framework via Self-Supervised Contrastive Learning and Quantum-Enhanced Feature Modeling

Jingsong Xia, Siqi Wang

TL;DR

This work tackles annotation scarcity and limited computational resources in medical image classification by proposing a lightweight hybrid framework that combines SimCLR-style self-supervised contrastive learning with a quantum-enhanced feature module. It employs a two-stage training scheme: unlabeled data pretraining to learn a robust encoder, followed by supervised fine-tuning on a small labeled set. A parameterized quantum circuit is integrated as a feature transformer with residual fusion to enrich classical representations. On coronary angiography data, the SSL-Quantum framework achieves state-of-the-art performance with around 2–3 million parameters, demonstrating strong accuracy, AUC, and F1 stability under low-resource conditions.

Abstract

Intelligent medical image analysis is essential for clinical decision support but is often limited by scarce annotations, constrained computational resources, and suboptimal model generalization. To address these challenges, we propose a lightweight medical image classification framework that integrates self-supervised contrastive learning with quantum-enhanced feature modeling. MobileNetV2 is employed as a compact backbone and pretrained using a SimCLR-style self-supervised paradigm on unlabeled images. A lightweight parameterized quantum circuit (PQC) is embedded as a quantum feature enhancement module, forming a hybrid classical-quantum architecture, which is subsequently fine-tuned on limited labeled data. Experimental results demonstrate that, with only approximately 2-3 million parameters and low computational cost, the proposed method consistently outperforms classical baselines without self-supervised learning or quantum enhancement in terms of Accuracy, AUC, and F1-score. Feature visualization further indicates improved discriminability and representation stability. Overall, this work provides a practical and forward-looking solution for high-performance medical artificial intelligence under resource-constrained settings.

A Lightweight Medical Image Classification Framework via Self-Supervised Contrastive Learning and Quantum-Enhanced Feature Modeling

TL;DR

This work tackles annotation scarcity and limited computational resources in medical image classification by proposing a lightweight hybrid framework that combines SimCLR-style self-supervised contrastive learning with a quantum-enhanced feature module. It employs a two-stage training scheme: unlabeled data pretraining to learn a robust encoder, followed by supervised fine-tuning on a small labeled set. A parameterized quantum circuit is integrated as a feature transformer with residual fusion to enrich classical representations. On coronary angiography data, the SSL-Quantum framework achieves state-of-the-art performance with around 2–3 million parameters, demonstrating strong accuracy, AUC, and F1 stability under low-resource conditions.

Abstract

Intelligent medical image analysis is essential for clinical decision support but is often limited by scarce annotations, constrained computational resources, and suboptimal model generalization. To address these challenges, we propose a lightweight medical image classification framework that integrates self-supervised contrastive learning with quantum-enhanced feature modeling. MobileNetV2 is employed as a compact backbone and pretrained using a SimCLR-style self-supervised paradigm on unlabeled images. A lightweight parameterized quantum circuit (PQC) is embedded as a quantum feature enhancement module, forming a hybrid classical-quantum architecture, which is subsequently fine-tuned on limited labeled data. Experimental results demonstrate that, with only approximately 2-3 million parameters and low computational cost, the proposed method consistently outperforms classical baselines without self-supervised learning or quantum enhancement in terms of Accuracy, AUC, and F1-score. Feature visualization further indicates improved discriminability and representation stability. Overall, this work provides a practical and forward-looking solution for high-performance medical artificial intelligence under resource-constrained settings.
Paper Structure (19 sections, 14 equations, 4 figures, 1 table)

This paper contains 19 sections, 14 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Overall Framework and Two-Stage Training Paradigm
  4. Self-Supervised Contrastive Learning for Medical Imaging
  5. Medical Image-Oriented Contrastive Augmentation Mapping
  6. Geometric Consistency-Driven Contrastive Embedding Learning
  7. Lightweight Quantum-Enhanced Encoder Design
  8. Learnable Mapping from Classical Features to Quantum States
  9. Parameterized Quantum Variational Evolution with Residual Enhancement
  10. Supervised Fine-Tuning and Task-Specific Discriminative Modeling
  11. Dataset
  12. Results
  13. Overall Performance Comparison Across Models
  14. Comprehensive Multi-Metric Performance Visualization
  15. Analysis of Classification Behavior via Confusion Matrices
  16. ...and 4 more sections

Figures (4)

  • Figure 1: Key Performance Metrics for Models.
  • Figure 2: Multi-Dimensional Performance Metrics.
  • Figure 3: Comparison Across Baseline and Proposed Models.
  • Figure 4: ROC Curves and AUC for Comparative Model Assessment.