Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MEDIC-AD: Towards Medical Vision-Language Model's Clinical Intelligence

Woohyeon Park, Jaeik Kim, Sunghwan Steve Cho, Pa Hong, Wookyoung Jeong, Yoojin Nam, Namjoon Kim, Ginny Y. Wong, Ka Chun Cheung, Jaeyoung Do

Abstract

Lesion detection, symptom tracking, and visual explainability are central to real-world medical image analysis, yet current medical Vision-Language Models (VLMs) still lack mechanisms that translate their broad knowledge into clinically actionable outputs. To bridge this gap, we present MEDIC-AD, a clinically oriented VLM that strengthens these three capabilities through a stage-wise framework. First, learnable anomaly-aware tokens (<Ano>) encourage the model to focus on abnormal regions and build more discriminative lesion centered representations. Second, inter image difference tokens (<Diff>) explicitly encode temporal changes between studies, allowing the model to distinguish worsening, improvement, and stability in disease burden. Finally, a dedicated explainability stage trains the model to generate heatmaps that highlight lesion-related regions, offering clear visual evidence that is consistent with the model's reasoning. Through our staged design, MEDIC-AD steadily boosts performance across anomaly detection, symptom tracking, and anomaly segmentation, achieving state-of-the-art results compared with both closed source and medical-specialized baselines. Evaluations on real longitudinal clinical data collected from real hospital workflows further show that MEDIC-AD delivers stable predictions and clinically faithful explanations in practical patient-monitoring and decision-support workflows

MEDIC-AD: Towards Medical Vision-Language Model's Clinical Intelligence

Abstract

Lesion detection, symptom tracking, and visual explainability are central to real-world medical image analysis, yet current medical Vision-Language Models (VLMs) still lack mechanisms that translate their broad knowledge into clinically actionable outputs. To bridge this gap, we present MEDIC-AD, a clinically oriented VLM that strengthens these three capabilities through a stage-wise framework. First, learnable anomaly-aware tokens (<Ano>) encourage the model to focus on abnormal regions and build more discriminative lesion centered representations. Second, inter image difference tokens (<Diff>) explicitly encode temporal changes between studies, allowing the model to distinguish worsening, improvement, and stability in disease burden. Finally, a dedicated explainability stage trains the model to generate heatmaps that highlight lesion-related regions, offering clear visual evidence that is consistent with the model's reasoning. Through our staged design, MEDIC-AD steadily boosts performance across anomaly detection, symptom tracking, and anomaly segmentation, achieving state-of-the-art results compared with both closed source and medical-specialized baselines. Evaluations on real longitudinal clinical data collected from real hospital workflows further show that MEDIC-AD delivers stable predictions and clinically faithful explanations in practical patient-monitoring and decision-support workflows

Paper Structure

This paper contains 36 sections, 4 equations, 6 figures, 10 tables.

Table of Contents

  1. Introduction
  2. RQ1: How can lesion and symptom recognition be improved in VLMs for real clinical settings?
  3. RQ2: How can a VLM disentangle temporal medical images to enable more accurate symptom tracking?
  4. RQ3: How can visual explainability be integrated into medical VLM reasoning?
  5. Related Works
  6. Vision-Language Models for Medical Imaging
  7. Zero-Shot Anomaly Detection
  8. Explainability in Vision--Language Models
  9. Methodology
  10. Overview
  11. Architecture and Stage-wise Training
  12. Stage 1: Anomaly Detection.
  13. Stage 2: Difference Reasoning.
  14. Stage 3: Visual Explainability.
  15. Experiments
  16. ...and 21 more sections

Figures (6)

  • Figure 1: Overall performance of VLMs on Medical Anomaly Detection and Medical Symptom Tracking (MMXU mmxu).
  • Figure 2: Comparison of VLMs on clinical applications. Medic-AD provides stronger lesion detection, temporal reasoning, and visual grounding than GPT-4o gpt4o and Citrus-V citrus_v.
  • Figure 3: Architecture of Medic-AD. (a) Stage 1: <Ano> Token Generation, (b) Stage 2: <Diff> Token Generation, and (c) Stage 3: Heatmap Generation illustrate each stage of the proposed framework. Note that CA denotes Cross-Attention.
  • Figure 4: Visual Grounding comparison between Medic-AD and Citrus-V citrus_v on diverse abnormal and normal samples.
  • Figure 5: Hyperparameter sensitivity analysis on (a) query token pooling size and (b) visual soft prompt counts. The red line denotes the baseline performance of Lingshu lingshu.
  • ...and 1 more figures