New C2A model improves chest X-ray classification by coupling spatial and clinical data

Researchers have developed a new classification head called C$^2$A (Co-occurrence Aware Class Attention) designed to improve the accuracy of multi-label classification for chest X-rays. This method explicitly links spatial information from the X-ray images with prior clinical knowledge about how different thoracic pathologies tend to co-occur. By learning per-class spatial attention maps and coupling these with empirical label co-occurrence data, C$^2$A can better identify and differentiate findings, particularly those with ambiguous spatial evidence. The approach demonstrated superior performance on the CheXpert dataset, achieving a macro-mean AUROC of 0.895 and showing significant gains for classes that frequently appear together. AI

IMPACT This research could lead to more accurate and reliable AI-powered diagnostic tools for medical imaging.

RANK_REASON The cluster contains a research paper detailing a new model for medical image classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• atelectasis
• C$^2$A
• CheXpert
• Greedy Coordinate Gradient
• Nagur Shareef Shaik

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →