Dissolving Is Amplifying: Towards Fine-Grained Anomaly Detection

TL;DR

This paper introduces DIA, a novel framework that uses dissolving transformations and contrastive learning to enhance fine-grained anomaly detection in medical images, achieving state-of-the-art results.

Contribution

The paper proposes a new dissolving is amplifying framework that improves fine-grained anomaly detection by removing and emphasizing subtle features in medical images.

Findings

Achieves around 18.40% AUC boost over baseline methods.

Outperforms existing benchmark methods in fine-grained anomaly detection.

Provides a self-supervised approach for semantic representation learning.

Abstract

Medical imaging often contains critical fine-grained features, such as tumors or hemorrhages, crucial for diagnosis yet potentially too subtle for detection with conventional methods. In this paper, we introduce \textit{DIA}, dissolving is amplifying. DIA is a fine-grained anomaly detection framework for medical images. First, we introduce \textit{dissolving transformations}. We employ diffusion with a generative diffusion model as a dedicated feature-aware denoiser. Applying diffusion to medical images in a certain manner can remove or diminish fine-grained discriminative features. Second, we introduce an \textit{amplifying framework} based on contrastive learning to learn a semantically meaningful representation of medical images in a self-supervised manner, with a focus on fine-grained features. The amplifying framework contrasts additional pairs of images with and without dissolving transformations applied and thereby emphasizes the dissolved fine-grained features. DIA significantly improves the medical anomaly detection performance with around 18.40\% AUC boost against the baseline method and achieves an overall SOTA against other benchmark methods. Our code is available at \url{https://github.com/shijianjian/DIA.git}.