VLM Models and Automated Grading of Atopic Dermatitis

TL;DR

This paper evaluates the effectiveness of seven vision-language models in automatically grading the severity of atopic dermatitis from patient images, exploring their potential for explainable medical assessment.

Contribution

It presents a systematic evaluation of VLMs for dermatological grading, highlighting their capabilities and limitations in medical image analysis.

Findings

VLMs can assess AD severity with moderate accuracy.

Some models provide explainable insights into image features.

Performance varies significantly across different VLM architectures.

Abstract

The task of grading atopic dermatitis (or AD, a form of eczema) from patient images is difficult even for trained dermatologists. Research on automating this task has progressed in recent years with the development of deep learning solutions; however, the rapid evolution of multimodal models and more specifically vision-language models (VLMs) opens the door to new possibilities in terms of explainable assessment of medical images, including dermatology. This report describes experiments carried out to evaluate the ability of seven VLMs to assess the severity of AD on a set of test images.