SDPA++: A General Framework for Self-Supervised Denoising with Patch Aggregation

2025-10-19

AI Summary

OCT image denoising has long been hindered by the scarcity of paired clean-noisy training data. To address this, we propose a label-free self-supervised denoising framework that generates high-fidelity pseudo-ground-truths via self-fusion and enables end-to-end training on single noisy OCT images through patch aggregation and model ensemble. Our method eliminates reliance on paired data and introduces a multi-dimensional evaluation scheme—including Contrast-to-Noise Ratio (CNR), Modulation Transfer Function Similarity (MSR), Texture Preservation (TP), and Edge Preservation (EP)—to jointly optimize reconstruction quality and structural fidelity. Evaluated on real clinical OCT datasets, our approach significantly outperforms existing unsupervised and self-supervised methods: CNR improves by 12.6%, MSR by 9.3%, while preserving critical anatomical structures. The framework demonstrates strong interpretability and clinical applicability.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Segmentation
• Search and Optimization: Non-convex Optimization

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Security and Privacy: Data transparency and provenance

Abstract

Optical Coherence Tomography (OCT) is a widely used non-invasive imaging technique that provides detailed three-dimensional views of the retina, which are essential for the early and accurate diagnosis of ocular diseases. Consequently, OCT image analysis and processing have emerged as key research areas in biomedical imaging. However, acquiring paired datasets of clean and real-world noisy OCT images for supervised denoising models remains a formidable challenge due to intrinsic speckle noise and practical constraints in clinical imaging environments. To address these issues, we propose SDPA++: A General Framework for Self-Supervised Denoising with Patch Aggregation. Our novel approach leverages only noisy OCT images by first generating pseudo-ground-truth images through self-fusion and self-supervised denoising. These refined images then serve as targets to train an ensemble of denoising models using a patch-based strategy that effectively enhances image clarity. Performance improvements are validated via metrics such as Contrast-to-Noise Ratio (CNR), Mean Square Ratio (MSR), Texture Preservation (TP), and Edge Preservation (EP) on the real-world dataset from the IEEE SPS Video and Image Processing Cup. Notably, the VIP Cup dataset contains only real-world noisy OCT images without clean references, highlighting our method's potential for improving image quality and diagnostic outcomes in clinical practice.

Problem

• Addresses denoising OCT images without clean reference data
• Reduces speckle noise using self-supervised patch aggregation
• Enhances clinical image quality for accurate disease diagnosis

Innovation

• Self-supervised denoising framework using only noisy images
• Generates pseudo-ground-truth via self-fusion and denoising
• Trains ensemble models with patch-based aggregation strategy