Resource-efficient medical image classification for edge devices

2025-12-19

AI Summary

To address the deployment challenges of medical image classification models on resource-constrained edge devices, this paper proposes a synergistic optimization framework integrating quantization-aware training (QAT) and post-training quantization (PTQ), coupled with lightweight network adaptation and edge inference acceleration. Evaluated on multimodal datasets including X-ray and MRI, the method achieves 4–8× model size reduction, 3–5× latency reduction, and less than 1.2% accuracy degradation—remaining within clinically acceptable thresholds. The core innovation lies in the deep coupling of quantization strategies with medical imaging characteristics: a lesion-sensitive region preservation mechanism and modality-adaptive calibration jointly ensure accuracy retention and deployment efficiency. This significantly enhances the feasibility of real-time AI-assisted diagnosis in primary-care and telemedicine settings.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Medical and Biological Imaging
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Medical image classification is a critical task in healthcare, enabling accurate and timely diagnosis. However, deploying deep learning models on resource-constrained edge devices presents significant challenges due to computational and memory limitations. This research investigates a resource-efficient approach to medical image classification by employing model quantization techniques. Quantization reduces the precision of model parameters and activations, significantly lowering computational overhead and memory requirements without sacrificing classification accuracy. The study focuses on the optimization of quantization-aware training (QAT) and post-training quantization (PTQ) methods tailored for edge devices, analyzing their impact on model performance across medical imaging datasets. Experimental results demonstrate that quantized models achieve substantial reductions in model size and inference latency, enabling real-time processing on edge hardware while maintaining clinically acceptable diagnostic accuracy. This work provides a practical pathway for deploying AI-driven medical diagnostics in remote and resource-limited settings, enhancing the accessibility and scalability of healthcare technologies.

Problem

• Develops resource-efficient medical image classification for edge devices
• Optimizes quantization techniques to reduce computational and memory demands
• Enables real-time AI diagnostics in remote, resource-limited healthcare settings

Innovation

• Model quantization reduces computational overhead and memory requirements
• Quantization-aware training and post-training quantization optimize edge device performance
• Quantized models enable real-time processing with maintained diagnostic accuracy