See More, Change Less: Anatomy-Aware Diffusion for Contrast Enhancement

2025-12-08

AI Summary

Existing medical image contrast enhancement methods often distort anatomical structures, introduce artifacts, or miss small lesions due to the absence of explicit anatomical structure modeling and contrast dynamics characterization. To address this, we propose a structure-aware diffusion model that jointly incorporates anatomical priors and multi-phase relative contrast dynamics modeling, enabling selective enhancement of clinically critical regions while preserving anatomical fidelity—achieved via a registration-free end-to-end training paradigm. Key innovations include a structure-aware supervised loss, a registration-free multi-phase learning framework, and a unified inference scheme. Evaluated on six external CT datasets, our method significantly outperforms state-of-the-art approaches: SSIM increases by 14.2%, PSNR by 20.6%, and FID decreases by 50%; notably, cancer detection F1-score improves by 10% on non-contrast-enhanced CT scans. The method thus achieves superior visual quality and diagnostic utility simultaneously.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Structured Learning
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Image enhancement improves visual quality and helps reveal details that are hard to see in the original image. In medical imaging, it can support clinical decision-making, but current models often over-edit. This can distort organs, create false findings, and miss small tumors because these models do not understand anatomy or contrast dynamics. We propose SMILE, an anatomy-aware diffusion model that learns how organs are shaped and how they take up contrast. It enhances only clinically relevant regions while leaving all other areas unchanged. SMILE introduces three key ideas: (1) structure-aware supervision that follows true organ boundaries and contrast patterns; (2) registration-free learning that works directly with unaligned multi-phase CT scans; (3) unified inference that provides fast and consistent enhancement across all contrast phases. Across six external datasets, SMILE outperforms existing methods in image quality (14.2% higher SSIM, 20.6% higher PSNR, 50% better FID) and in clinical usefulness by producing anatomically accurate and diagnostically meaningful images. SMILE also improves cancer detection from non-contrast CT, raising the F1 score by up to 10 percent.

Problem

• Enhances medical images without distorting anatomical structures
• Improves contrast in clinically relevant regions only
• Boosts cancer detection accuracy from non-contrast CT scans

Innovation

• Anatomy-aware diffusion model for medical contrast enhancement
• Registration-free learning from unaligned multi-phase CT scans
• Unified inference for fast, consistent enhancement across phases