搜索医学 AI 论文与资源

ICLR 2026 Poster accepted paper at ICLR 2026. Current methods for detecting spurious correlations rely on data splits or error patterns, leaving many harmful shortcuts invisible when counterexamples are absent. We introduce BEE (Bridging Explainability and Embeddings), a framework that shifts the focus from model predictions to the weight space and embedding geometry underlying decisions. By analyzing how fine-tuning perturbs pretrained representations, BEE uncovers spurious correlations that remain hidden from conventional evaluation pipelines. We use linear probing as a transparent diagnostic lens, revealing spurious features that not only persist after full fine-tuning but also transfer across diverse state-of-the-art models. Code/project link: https://github.com/bit-ml/bee

ICLR 2026 Poster accepted paper at ICLR 2026. Visual Question Answering (VQA) enables targeted and context-dependent analysis of medical images, such as chest X-rays (CXRs). However, existing VQA datasets for CXRs are typically constrained by simplistic and brief answer formats, lacking localization annotations (e.g., bounding boxes) and structured tags (e.g., region or radiological finding/disease tags). To address these limitations, we introduce MIMIC-Ext-CXR-QBA (abbr. CXR-QBA), a large-scale CXR VQA dataset derived from MIMIC-CXR, comprising 42 million QA-pairs with multi-granular, multi-part answers, detailed bounding boxes, and structured tags. Code/project link: https://github.com/philip-mueller/mimic-ext-cxr-qba/

ICLR 2026 Poster accepted paper at ICLR 2026. Vision-Language Models (VLMs) for radiology report generation are typically trained to mimic the narrative flow of human experts. However, we identify a potential limitation in this conventional paradigm. We hypothesize that optimizing for narrative coherence encourages models to rely on linguistic priors and inter-sentence correlations, which can weaken their grounding in direct visual evidence and lead to factual inaccuracies. To investigate this, we design a controlled experiment demonstrating that as textual context increases, a model's reliance on the input image systematically decays. We propose LLaVA-TA (Topic-guided and Anatomy-aware), a new fine-tuning framework that directly addresses this challenge by re-engineering the generation process.

The RSNA Pneumonia Detection Challenge dataset is a chest radiograph benchmark for detecting pneumonia-related lung opacities. It supports object detection, chest X-ray classification, localization, and radiology AI evaluation under a competition framework.

VinDr-CXR is a chest X-ray dataset with radiologist annotations from Vietnamese hospitals. It supports abnormality classification, lesion localization, radiology object detection, and robustness studies across clinical sites and populations.

NIH ChestX-ray14 is a public chest radiograph dataset with image-level labels for thoracic disease findings mined from reports. It is commonly used for chest X-ray classification, weak supervision, thoracic disease detection, and radiology benchmark comparisons.

CheXpert is a large chest radiograph dataset from Stanford with uncertainty-aware labels for common chest X-ray findings. It is widely used for radiology classification, label uncertainty modeling, chest X-ray representation learning, and clinical imaging benchmarks.

MIMIC-CXR is a large deidentified chest radiograph dataset with associated free-text radiology reports. It is widely used for chest X-ray classification, report generation, image-text representation learning, radiology retrieval, and medical multimodal foundation model evaluation.

JPG-formatted chest radiographs with labels derived from free-text reports, hosted by PhysioNet.