Medical vision–language pretraining increasingly relies on medical reports as large-scale supervisory signals; however, raw reports often exhibit substantial stylistic heterogeneity, variable length, and a considerable amount of image-irrelevant content. Although text normalization is frequently adopted as a preprocessing step in prior work, its design principles and empirical impact on vision-language pretraining remain insufficiently and systematically examined. In this study, we present MedTri, a deployable normalization framework for medical vision-language pretraining that converts free-text reports into a unified [Anatomical Entity: Radiologic Description + Diagnosis Category] triplet. This structured, anatomy-grounded normalization preserves essential morphological and spatial information while removing stylistic noise and image-irrelevant content, providing consistent and image-grounded textual supervision at scale. Across multiple datasets spanning both X-ray and computed tomography (CT) modalities, we demonstrate that structured, anatomy-grounded text normalization is an important factor in medical vision–language pretraining quality, yielding consistent improvements over raw reports and existing normalization baselines. In addition, we illustrate how this normalization can easily support targeted text-level augmentation strategies, including knowledge enrichment and anatomy-grounded counterfactual supervision, which provide complementary gains in robustness and generalization without altering the core normalization process. Together, our results position structured text normalization as a critical and generalizable preprocessing component for medical vision–language learning, and establish MedTri as a practical implementation of this approach. The source code is available at https://github.com/Arturia-Pendragon-Iris/MedTri .
Long-term retrospective medical imaging archives cannot be directly utilized due to the lack of fine-grained annotations, requiring extensive manual curation by expert physicians in most previous medical imaging studies. Furthermore, the absence of standardized archiving protocols and the heavy reliance on subjective judgment for image retention lead to significant variations in the number of images per patient, while the presence of irrelevant frames introduces substantial noise. We propose FARSIGHT, an end-to-end archive-to-prediction framework designed to overcome these universal bottlenecks, evaluated here on renal transplant chronic allograft injury assessment. FARSIGHT integrates the Modality Aware Governance and Optimization System (MAGOS), an offline vision-language-model-guided curation agent, with hierarchical multi-modal multi-instance learning (HMMIL) for patient-level prediction. On a retrospective cohort of 4,576 patients comprising 41,203 images across four distinct modalities, MAGOS achieves 0.999 routing accuracy with complete parsing success. Using these curated inputs, HMMIL achieves a test AUC of 0.816 and an accuracy of 0.746 for patient-level prediction, while handling variable-length studies and missing modalities. The code is available at: https://github.com/hanlliumu/Farsight .
Medical foundation models are increasingly used as generic visual backbones for downstream diagnosis, yet existing fairness studies in medical imaging mostly focus on a single protected attribute (e.g., age or sex) and rarely examine intersectional subgroups defined by multiple demographics. Compared to single-attribute settings, multi-attribute fairness introduces additional challenges: (i) intersectional subgroup data sparsity, where some demographic combinations contain very few samples, making subgroup-specific adaptation unstable and prone to overfitting; and (ii) sensitive attribute leakage (proxy discrimination), where learned representations retain correlated demographic information and allow models to rely on demographic shortcuts rather than clinically relevant visual features. To address this, we propose Fair Gated-Expert adaptation with multi-attribute gating (FairGE), a fairness-aware adaptation framework for medical foundation models. FairGE integrates two complementary components: a Multi-attribute Gated distribution-aware Mixture-of-Experts (MA-dMoE) module that employs shared experts with multi-attribute gating to enable intersectional adaptation while avoiding overfitting on small subgroups, and a Multi-attribute Purification (MAP) module that combines a lightweight MLP with multiple adversarial attribute classifiers (one for each sensitive attribute) trained via gradient reversal to prevent the model from encoding sensitive demographic information in the final representation. Experiments on two representative medical benchmarks, HAM10000 and Harvard-FairVLMed, show that FairGE consistently reduces both marginal and intersectional disparities, improves worst-group performance, and preserves overall AUC. Code is available at https://github.com/yuchenshao749/FairGE .
Radiology report generation (RRG) increasingly relies on rich evidence, most commonly retrieved similar reports from other patients and longitudinal history from the same patient. Prior approaches often concatenate evidence into a long prompt, which can degrade performance. Such long contexts are typically noisy and only partially relevant, making it challenging for the backbone decoder to reliably select the most relevant evidence for the current query. We propose HM-RRG, a hierarchical-memory framework that organizes and selects evidence instead of treating all context as a flat sequence. HM-RRG builds an image-conditioned retrieval memory bank by compressing each retrieved report into a compact memory vector grounded on the current image. It then processes the patient’s prior reports as a chronological stream, maintains segment-level longitudinal memories, and performs hierarchical retrieval over the union of cross-patient and within-patient memories to form an evidence prompt for each segment during decoding. This design enables selective recall of relevant information while suppressing irrelevant or conflicting details under long contexts. Experiments on MIMIC-CXR with Longitudinal-MIMIC show that HM-RRG improves both text quality and clinical accuracy. In particular, HM-RRG achieves the best factuality metrics among state-of-the-art RRG models, demonstrating that structured memory is effective for radiology report generation. Our code is available at https://github.com/QingyueJ-nd/HM-RRG .
