# How Medical VLMs Underutilize Their Vision Encoders: A Dermatology Perspective

arXiv: https://arxiv.org/abs/2609.36557 · HTML: https://mlsanity.com/paper/2609.36557/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 32 of its 48 references that are in the map.
- First posted on arXiv 29 Sep 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Medical vision-language models](https://mlsanity.com/mllm/medical-vision-language-models/)

## Builds on (most cited first, 30 of 32)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison](https://mlsanity.com/paper/1901.07031/) (2019)
- [Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs](https://mlsanity.com/paper/2401.06209/) (2024)
- [MedMNIST v2 -- A large-scale lightweight benchmark for 2D and 3D biomedical image classification](https://mlsanity.com/paper/2110.14795/) (2021)
- [Rotation Equivariant CNNs for Digital Pathology](https://mlsanity.com/paper/1806.03962/) (2018)
- [Visual Sketchpad: Sketching as a Visual Chain of Thought for Multimodal Language Models](https://mlsanity.com/paper/2406.09403/) (2024)
- [PMC-CLIP: Contrastive Language-Image Pre-training using Biomedical Documents](https://mlsanity.com/paper/2303.07240/) (2023)
- [BiomedGPT: A Generalist Vision-Language Foundation Model for Diverse Biomedical Tasks](https://mlsanity.com/paper/2305.17100/) (2023)
- [Evaluating Deep Neural Networks Trained on Clinical Images in Dermatology with the Fitzpatrick 17k Dataset](https://mlsanity.com/paper/2104.09957/) (2021)
- [Investigating the Catastrophic Forgetting in Multimodal Large Language Models](https://mlsanity.com/paper/2309.10313/) (2023)
- [A deep learning system for differential diagnosis of skin diseases](https://mlsanity.com/paper/1909.05382/) (2019)
- [PAD-UFES-20: a skin lesion dataset composed of patient data and clinical images collected from smartphones](https://mlsanity.com/paper/2007.00478/) (2020)
- [Look-Back: Implicit Visual Re-focusing in MLLM Reasoning](https://mlsanity.com/paper/2507.03019/) (2025)
- [A Multimodal Vision Foundation Model for Clinical Dermatology](https://mlsanity.com/paper/2410.15038/) (2024)
- [In-context learning enables multimodal large language models to classify cancer pathology images](https://mlsanity.com/paper/2403.07407/) (2024)
- [DPCore: Dynamic Prompt Coreset for Continual Test-Time Adaptation](https://mlsanity.com/paper/2406.10737/) (2024)
- [Re-Align: Aligning Vision Language Models via Retrieval-Augmented Direct Preference Optimization](https://mlsanity.com/paper/2502.13146/) (2025)
- [Derm1M: A Million-scale Vision-Language Dataset Aligned with Clinical Ontology Knowledge for Dermatology](https://mlsanity.com/paper/2503.14911/) (2025)
- [OT-VP: Optimal Transport-guided Visual Prompting for Test-Time Adaptation](https://mlsanity.com/paper/2407.09498/) (2024)
- [Automated Structured Radiology Report Generation](https://mlsanity.com/paper/2505.24223/) (2025)
- [MM-Skin: Enhancing Dermatology Vision-Language Model with an Image-Text Dataset Derived from Textbooks](https://mlsanity.com/paper/2505.06152/) (2025)
- [On Domain-Adaptive Post-Training for Multimodal Large Language Models](https://mlsanity.com/paper/2411.19930/) (2024)
- [Achieving Reliable and Fair Skin Lesion Diagnosis via Unsupervised Domain Adaptation](https://mlsanity.com/paper/2307.03157/) (2023)
- [Doctor Approved: Generating Medically Accurate Skin Disease Images through AI-Expert Feedback](https://mlsanity.com/paper/2506.12323/) (2025)
- [Specialized curricula for training vision-language models in retinal image analysis](https://mlsanity.com/paper/2407.08410/) (2024)
- [Prime Once, then Reprogram Locally: An Efficient Alternative to Black-Box Service Model Adaptation](https://mlsanity.com/paper/2604.01474/) (2026)

## Abstract (arXiv)

Medical Vision-Language Models (VLMs) show significant promise for clinical image understanding, offering accurate diagnosis with interpretable reasoning. However, a critical performance gap exists between their strong vision encoders and the full multimodal model: in dermatology, the MedSigLIP encoder outperforms MedGemma by an average of 10.26 percentage points even when both use zero target-task labels; few-shot linear probing provides further evidence of strong visual representations. This gap motivates an investigation of how visual information is used in end-to-end diagnosis and why plausible-sounding predictions can lack grounding in image evidence. Using dermatology as our primary testbed, we systematically investigate three hypotheses for this phenomenon. We further provide a mechanistic analysis of the model's internal attention patterns, showing that a simple describe-then-decide prompting strategy increases vision attention by 30-40% during generation. Task-specific fine-tuning improves dermatology classification but reduces cross-domain medical question-answering performance in our evaluation. To address these challenges, we combine label-free prompting with low-label encoder-assisted reranking while keeping the VLM frozen. We validate the interventions across five VLM backbones in dermatology and provide supporting representation and attention analyses across additional medical modalities.