X-ray Insights Unleashed: Pioneering the Enhancement of Multi-Label Long-Tail Data

2025-12-24

AI Summary

Addressing the challenge of diagnosing rare pulmonary abnormalities in chest X-rays—exacerbated by multi-label long-tailed label distributions—this paper proposes a normal-X-ray-driven diffusion model synthesis framework. Our method leverages a diffusion model pre-trained on large-scale normal chest radiographs and performs lesion-region-controllable inpainting to augment tail-class samples. To enhance fine-tuning stability and semantic consistency, we introduce two novel components: (i) Large Language Model-guided Knowledge injection (LKG), which incorporates clinical priors into the diffusion process, and (ii) Progressive Incremental Learning (PIL), which mitigates catastrophic forgetting during long-tail adaptation. Evaluated on MIMIC-CXR and CheXpert, our approach significantly improves recognition performance for tail classes, establishing new state-of-the-art results for multi-label long-tailed medical image classification. The framework offers a scalable, interpretable paradigm for few-shot diagnosis of rare pathologies.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• 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
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Long-tailed pulmonary anomalies in chest radiography present formidable diagnostic challenges. Despite the recent strides in diffusion-based methods for enhancing the representation of tailed lesions, the paucity of rare lesion exemplars curtails the generative capabilities of these approaches, thereby leaving the diagnostic precision less than optimal. In this paper, we propose a novel data synthesis pipeline designed to augment tail lesions utilizing a copious supply of conventional normal X-rays. Specifically, a sufficient quantity of normal samples is amassed to train a diffusion model capable of generating normal X-ray images. This pre-trained diffusion model is subsequently utilized to inpaint the head lesions present in the diseased X-rays, thereby preserving the tail classes as augmented training data. Additionally, we propose the integration of a Large Language Model Knowledge Guidance (LKG) module alongside a Progressive Incremental Learning (PIL) strategy to stabilize the inpainting fine-tuning process. Comprehensive evaluations conducted on the public lung datasets MIMIC and CheXpert demonstrate that the proposed method sets a new benchmark in performance.

Problem

• Enhances rare lung anomaly detection in chest X-rays
• Augments tail lesion data using normal X-ray diffusion models
• Integrates language model guidance to stabilize inpainting training

Innovation

• Use normal X-rays to train diffusion model for data synthesis
• Inpaint head lesions to preserve tail classes as augmented data
• Integrate LLM guidance with incremental learning for stable fine-tuning