Learning Sparse Label Couplings for Multilabel Chest X-Ray Diagnosis

2025-11-11

AI Summary

This paper addresses key challenges in multi-label chest X-ray classification—extreme label imbalance, asymmetric misdiagnosis costs, and inadequate modeling of label co-occurrence—by proposing a lightweight, efficient diagnostic framework. The core innovation is a learnable sparse label graph refinement module that explicitly captures label dependencies at the logits level via single-step message passing, introducing negligible computational overhead. The method integrates a SE-ResNeXt101 backbone with asymmetric loss, mixed-precision training, cosine annealing, and exponential moving average (EMA) for robust optimization, and employs multi-fold cross-validation with test-time augmentation (TTA) ensembling. Experiments demonstrate significant improvement in macro-AUC, achieving high performance without additional annotations. The framework is computationally efficient, hardware-friendly, and clinically deployable.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We study multilabel classification of chest X-rays and present a simple, strong pipeline built on SE-ResNeXt101 $(32 imes 4d)$. The backbone is finetuned for 14 thoracic findings with a sigmoid head, trained using Multilabel Iterative Stratification (MIS) for robust cross-validation splits that preserve label co-occurrence. To address extreme class imbalance and asymmetric error costs, we optimize with Asymmetric Loss, employ mixed-precision (AMP), cosine learning-rate decay with warm-up, gradient clipping, and an exponential moving average (EMA) of weights. We propose a lightweight Label-Graph Refinement module placed after the classifier: given per-label probabilities, it learns a sparse, trainable inter-label coupling matrix that refines logits via a single message-passing step while adding only an L1-regularized parameter head. At inference, we apply horizontal flip test-time augmentation (TTA) and average predictions across MIS folds (a compact deep ensemble). Evaluation uses macro AUC averaging classwise ROC-AUC and skipping single-class labels in a fold to reflect balanced performance across conditions. On our dataset, a strong SE-ResNeXt101 baseline attains competitive macro AUC (e.g., 92.64% in our runs). Adding the Label-Graph Refinement consistently improves validation macro AUC across folds with negligible compute. The resulting method is reproducible, hardware-friendly, and requires no extra annotations, offering a practical route to stronger multilabel CXR classifiers.

Problem

• Addresses multilabel classification of chest X-rays for thoracic findings
• Handles extreme class imbalance and asymmetric error costs in diagnosis
• Learns sparse inter-label couplings to refine predictions efficiently

Innovation

• Fine-tuned SE-ResNeXt101 backbone for thoracic findings
• Proposed lightweight Label-Graph Refinement with sparse coupling
• Employed asymmetric loss and compact deep ensemble strategy