Synergistic Perception-Reasoning Governance: Grounding Medical MLLMs with Verifiable Anatomical Evidence

Counterfactual Anatomy-guided Spatial-Temporal Decoding for Annotation-Free Hallucination Mitigation in Medical VLMs

Experiments on the SLAKE and MIMIC-CXR datasets demonstrate that CAST consistently outperforms strong baselines and surpasses decoding strategies reliant on ground truth, and indicates that compact, automatically selected regions provide highly effective contrastive guidance without expert annotations.

Post-Hoc Explainability for Generative Volumetric Medical Vision-Language Models

In radiology, multimodal vision-language models (VLMs) are increasingly used for clinical tasks such as report generation and visual question answering. Their adoption has raised important concerns regarding the transparency and trust-worthiness of the generated text reports, as the reasoning process leading to a given report remains opaque for clinicians. In this work, we investigate how post-hoc attribution methods behave in large generative volumetric VLMs that jointly process 3D scans and clinical text, a setting that remains largely unexplored. We propose to use Post-hoc gradient-based attribution that directly links small changes in the input volume to changes in the model's output probability. The faithfulness and spatial specificity of four attribution methods are subsequently characterized and compared when applied to Med3DVLM, a recent medical VLM for image-text understanding. To enable differentiable attribution at inference time, we apply teacher forcing exclusively during the attribution forward pass so that the stochastic generation path is converted into a differentiable computation graph. The cumulative log-likelihood of the generated response is used as the scalar attribution target. Input-level methods, including Saliency and Integrated Gradients, alongside feature-level techniques such as Grad-CAM and Guided Grad-CAM, are computed on volumetric inputs across diverse clinical question types. Qualitative assessment and a quantitative voxel deletion protocol indicate that input-space gradient methods, especially Integrated Gradients, produce spatially selective and causally faithful relevance maps, whereas feature-level methods like Layer Grad-CAM exhibit significant spatial diffusion and fail to provide discriminative localization in multimodal architectures. These results demonstrate that input-level attributions exhibit a significantly steeper drop in model confidence compared to feature-level methods, confirming they are more faithful and spatially precise, while intermediate-layer maps remain diffuse and non-discriminative.

MedVCoT: Bridging the Modality Gap in Medical VQA Through Latent Visual Reasoning

MedQwen-VGR1: Multimodal Vision-Guided Reasoning Model for Med-VQA, Temporal Diagnosis and Drug Interaction Analysis in Medical Imaging and Pathology

MedQwen-VGR1 is a novel multimodal visionlanguage model (VLM) trained for medical visual question answering (Med-VQA), longitudinal temporal diagnosis, and drug interaction analysis across radiology and pathology imagery. The model undergoes a multi-stage training pipeline comprising: (1) Continuous Domain-Adaptive Pretraining on heterogeneous medical image-text corpora; (2) Supervised FineTuning on expert-annotated multimodal datasets spanning sequential imaging conversations; (3) Human Preference Alignment via Direct Preference Optimization (DPO) and Group Relative Preference Optimization (GRPO) applied to Vision-Guided Chain-of-Thought (CoT) Reasoning trajectories; and (4) Smart Memory Module integration enables Cross-Visit Context Retention and Comparative Reasoning on sequential or temporal pathologies. MedQwen-VGR1 processes sequential medical images and textual queries to generate stepwise, evidence-grounded rationales correlating visual features across timepoints, predict adverse drug interactions from images/metadata, and output calibrated reliability scores. Evaluations on PathVQA, SLAKE, and VQA-RAD demonstrate superior performance over LLaVA-Med (+14.7% accuracy) and BioViL-T (+12.3% temporal reasoning), achieving 82.4% VQA accuracy and 78.6% change detection precision, enabling multi-turn clinical dialogues and automated clinical reports. The framework enables dynamic, multi-turn clinical dialogues with automated medical report synthesis, addressing critical gaps in temporal reasoning, explainability, and clinical safety alignment for realworld AI and safety enhanced clinical decision support systems.

RadSight: Towards Perceptually Reliable Multimodal Radiology Image Understanding

Medical multimodal large language models (MLLMs) are increasingly expected to perform complex image understanding tasks, yet their reliability is often compromised by frequent errors in visual interpretation. To systematically trace these failures, we traverse the hierarchy from high-level clinical tasks down to fundamental visual perception. We therefore introduce Perception-Bench, a large-scale benchmark comprising 1.13 million samples that assesses medical MLLMs across six dimensions: attribute judgment, spatial grounding, spatial understanding, disease prediction, anomaly detection, and report generation, spanning both 2D and 3D radiology images. Our analysis on Perception-Bench reveals that existing MLLMs lack the ability to capture even the most basic lesion attributes, such as location, size, and density. This inability to ground clinical outputs in primary visual evidence reveals that the models'diagnostic unreliability is rooted in a critical but overlooked bottleneck in low-level visual perception. Motivated by this, we propose RadSight, a perception-driven MLLM built upon a dual 2D/3D encoder architecture that preserves native imaging spatial structures. RadSight formulates medical image understanding as a four-stage progressive process: visual-language alignment, fine-grained visual perception, clinical diagnosis, and diagnostic interpretation. The model is trained on an 8.37 million perception-oriented corpus using progressive curriculum learning. On Perception-Bench, RadSight consistently outperforms existing MLLMs across all six evaluation dimensions, with particularly strong gains in spatial grounding and clinical diagnosis. It also achieves consistent improvements on public 2D and 3D medical benchmarks, further demonstrating that robust low-level visual perception is a critical foundation for reliable clinical understanding. Code and model will be publicly available.

Med-OPD: Improving Medical Vision-Language Models via Evidence-Aware On-Policy Distillation

Medical Vision-Language Models (Med-VLMs) require reliable reasoning from fine-grained visual evidence, yet existing models can produce plausible clinical answers by relying on language priors or medical templates rather than truly attending to diagnosis-critical regions. On-Policy Distillation (OPD) offers dense token-level supervision on student-generated trajectories and provides a privacy-compatible means of capability transfer without requiring the redistribution of raw patient data. However, standard OPD uniformly distills all tokens, causing sparse evidence-dependent tokens to be diluted by abundant clinical narrative tokens. Inspired by the success of OPD in the large language model community, we propose \textbf{Med-OPD}, to our knowledge the first unified post-training framework that integrates on-policy distillation with medical evidence-aware supervision for Med-VLMs. We introduce \textbf{Medical Evidence Advantage} (MEA), a teacher-grounded counterfactual signal that uses an answer-aware hint to focus teacher scoring on evidence supporting the target diagnosis, and measures each token's dependence on medical visual evidence by comparing teacher likelihoods under the original and evidence-degraded imaging modalities. Based on MEA, Med-OPD redistributes the distillation signal at both the token and trajectory levels, emphasizing diagnosis-critical tokens and evidence-reliant rollouts. Experiments on OmniMedVQA subsets show that Med-OPD consistently outperforms SFT and standard OPD across CT, MRI, Disease Diagnosis, and Lesion Grading. These results demonstrate that evidence-aware distillation can better strengthen medical VLMs'reliance on key visual evidence and improve reliable multimodal medical reasoning. The source code and data is publicly available at: https://github.com/yunhang8658/MedOPD.git