PLRS-IC: A Dual-Calibration Framework for Chest X-Ray Vision-Language Alignment

2026-09-30

AI Summary

This study addresses projection mismatch and cross-patient semantic overlap in chest X-ray vision-language alignment by proposing the PLRS-IC dual-calibration framework. This work introduces a novel dual-calibration mechanism to tackle these challenges: it mitigates projection discrepancies through locally projected conditional low-rank residual similarity and suppresses soft false negatives via global information content calibration. Furthermore, the framework optimizes contrastive learning by integrating parameter-efficient low-rank residuals with information-theoretic priors. Experimental results demonstrate that the proposed method significantly improves zero-shot classification, localization, and segmentation performance across nine public benchmarks, effectively validating the superiority of the dual-calibration strategy.

Abstract

Fine-grained vision-language alignment in chest radiography enables zero-shot classification, grounding, and segmentation without task-specific annotations. However, this alignment is fundamentally hindered by two intertwined sources of ambiguity: projection-induced visual mismatch and patient-agnostic semantic overlap. First, at the local feature level, frontal and lateral radiographs exhibit distinct appearances for the same clinical finding, rendering a shared patch-text similarity geometry inherently suboptimal. Compounding this visual ambiguity is a semantic mismatch during global contrastive optimization, where instance-level objectives penalize cross-patient pairs as strict negatives even when they share identical positive clinical concepts. To address this dual ambiguity, we propose PLRS-IC, a unified dual-calibration framework for chest X-ray representation learning. At the local alignment stage, Projection-Conditioned Low-Rank Residual Similarity (PLRS) dynamically adapts patch-text matching to projection-specific manifolds using a bounded, parameter-efficient low-rank residual. At the global optimization stage, Information-Content-Calibrated Soft False-Negative Suppression (IC-SFNS) leverages a corpus-derived information-theoretic prior to soften the penalty of semantically overlapping negatives without altering original contrastive assignments. Extensive experiments across nine public zero-shot benchmark settings demonstrate that our framework yields consistent improvements in classification, grounding, and segmentation, validating the necessity of dual-calibration in medical vision-language pre-training.

Problem

• Vision-Language Alignment
• Chest X-Ray
• Projection Ambiguity
• Semantic Overlap
• Zero-Shot Learning

Innovation

• Vision-Language Alignment
• Dual-Calibration Framework
• Low-Rank Residual Similarity
• Soft False-Negative Suppression
• Zero-Shot Learning