Melan-Dx: a knowledge-enhanced vision-language framework improves differential diagnosis of melanocytic neoplasm pathology - Takeaways - MDSpire

Melan-Dx: a knowledge-enhanced vision-language framework improves differential diagnosis of melanocytic neoplasm pathology

  • By

  • Jialu Yao

  • Songhao Li

  • Peixian Liang

  • Xiaowei Xu

  • David Elder

  • Zhi Huang

  • January 20, 2026

  • 0 min

Share

  • 1

    Melanoma is a leading cause of skin cancer deaths and is often misdiagnosed due to the complexity of its histologic subtypes.

  • 2

    Melan-Dx is an AI framework that enhances differential diagnosis of melanocytic neoplasms using a curated image and knowledge database.

  • 3

    The dataset for Melan-Dx includes 2893 images and 1102 expert-annotated knowledge entries from dermatopathologists.

  • 4

    Melan-Dx achieved an accuracy of 0.869 for binary classification and significant improvements over existing diagnostic methods.

  • 5

    The framework allows for enhanced pathology model performance without the need for fine-tuning the vision backbone.

Original Source(s)

Melan-Dx: a knowledge-enhanced vision-language framework improves differential diagnosis of melanocytic neoplasm pathology

  • by Jialu Yao, Songhao Li, Peixian Liang, Xiaowei Xu, David Elder, Zhi Huang

  • January 20, 2026

Related Content

Blood Test Predicts Immunotherapy Response

AI-based analysis of blood DNA maps the complex ecosystem surrounding cancer cells

  • May 18, 2026

  • 3 min
Myelodysplastic syndromes complicated by atypical Sweet syndrome: a brief research report

Skin cancer risk in alopecia areata: a systematic review and meta-analysis