Progressive Transfer Learning for Multi-Pass Fundus Image Restoration

2025-04-14

AI Summary

Diabetic retinopathy (DR) early screening relies critically on high-quality fundus images; however, clinical acquisitions often suffer from degradation—including insufficient illumination, noise, blur, and motion artifacts—severely compromising the reliability of automated diagnosis. To address this, we propose a multi-round progressive blind restoration framework, introducing the first pairing-free iterative transfer learning paradigm: an initial coarse restoration via CycleGAN is followed by stage-wise fine-tuning and unsupervised domain adaptation to progressively refine reconstruction, enhancing structural fidelity while preserving diagnostically critical pathological features. Evaluated on the DeepDRiD dataset, our method achieves state-of-the-art performance (significantly improved PSNR and SSIM). Clinical validation confirms substantial enhancement in discriminability of microaneurysms, hemorrhages, and other subtle lesions. This work establishes a novel, robust paradigm for DR screening directly from low-quality fundus imagery.

Technology Category

• Computer Vision: Image and Video Retrieval
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Diabetic retinopathy is a leading cause of vision impairment, making its early diagnosis through fundus imaging critical for effective treatment planning. However, the presence of poor quality fundus images caused by factors such as inadequate illumination, noise, blurring and other motion artifacts yields a significant challenge for accurate DR screening. In this study, we propose progressive transfer learning for multi pass restoration to iteratively enhance the quality of degraded fundus images, ensuring more reliable DR screening. Unlike previous methods that often focus on a single pass restoration, multi pass restoration via PTL can achieve a superior blind restoration performance that can even improve most of the good quality fundus images in the dataset. Initially, a Cycle GAN model is trained to restore low quality images, followed by PTL induced restoration passes over the latest restored outputs to improve overall quality in each pass. The proposed method can learn blind restoration without requiring any paired data while surpassing its limitations by leveraging progressive learning and fine tuning strategies to minimize distortions and preserve critical retinal features. To evaluate PTL's effectiveness on multi pass restoration, we conducted experiments on DeepDRiD, a large scale fundus imaging dataset specifically curated for diabetic retinopathy detection. Our result demonstrates state of the art performance, showcasing PTL's potential as a superior approach to iterative image quality restoration.

Problem

• Enhancing degraded fundus images for reliable diabetic retinopathy screening
• Overcoming limitations of single-pass restoration with progressive transfer learning
• Achieving blind restoration without paired data while preserving retinal features

Innovation

• Progressive transfer learning for iterative enhancement
• Multi-pass restoration via Cycle GAN and PTL
• Unpaired blind restoration preserving retinal features