C$^2$A: Coupling Spatial Evidence with Clinical Priors via Co-occurrence Aware Class Attention for Multi-Label Chest X-Ray Classification
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors designed a new method called C²A to help computers better identify multiple lung diseases in X-ray images by paying attention to where diseases show up and how they often appear together. Their approach looks at specific areas linked to each disease and uses knowledge about which conditions usually happen together to improve guesses. They tested this on a dataset and found that C²A works better than earlier methods, especially for diseases that appear together and are hard to spot. This improvement comes with only a small increase in model complexity.
thoracic pathologiesmulti-label classificationspatial attentionlabel co-occurrencegraph neural networksmessage passingAUROCCheXpert datasetclinical priors
Authors
Akash Gogineni, Nagur Shareef Shaik, Aasrith Mandava, Adnan Masood, Dong Hye Ye
Abstract
Thoracic pathologies rarely occur in isolation, yet standard multi-label classifiers rely on shared global descriptors, discarding \emph{where} findings lie and \emph{how} they co-occur. We propose \textbf{C$\mathbf{^2}$A} (Co-occurrence Aware Class Attention), a classification head that explicitly couples spatial evidence with clinical priors. First, C$^2$A casts pooling as an expectation over learned per-class spatial attention maps, yielding localized descriptors for each disease. Second, it couples these descriptors via a learnable graph warm-started from empirical label co-occurrence. A single residual message-passing step shares evidence among related findings, proving to be a bounded perturbation of the identity where co-occurrence enters each logit through an explicit bilinear interaction. On CheXpert, C$^2$A achieves a superior $0.895$ macro-mean AUROC, outperforming advanced context-gating baselines. Crucially, gains concentrate on highly co-occurrent classes with ambiguous spatial evidence (rescuing Atelectasis by $+1.5$ over GCG), demonstrating the prior's regularizing effect with a negligible overhead of one linear projection and a $C\!\times\!C$ edge matrix.