We decompose the generalization gap of dermatology-AI models across skin tones, disentangling the effect of disease-distribution shift from skin-tone representation to explain why models underperform on underrepresented populations. Accepted at the MIRASOL Workshop, MICCAI 2026.
@article{kunwor2026dermatology,title={Disease Burden over Skin Tone: Decomposing the Dermatology-AI Generalization Gap},author={Kunwor, Nirajan and Poudel, Sanjaya and Trinh, Quoc-Huy and Arafat, Jahidul and Gaire, Sunil Kumar},journal={arXiv preprint arXiv:2609.02111},year={2026},}
DeCaF@MICCAI
Federated LoRA Adaptation of BiomedCLIP Across Four International Chest X-Ray Cohorts
Sanjaya Poudel, Nirajan Kunwor, Manish Dhakal, and 2 more authors
We study federated, parameter-efficient (LoRA) adaptation of BiomedCLIP across four international chest X-ray cohorts, evaluating aggregation strategies for robust cross-institution model adaptation without centralizing patient data. Accepted at the DeCaF Workshop, MICCAI 2026.
@article{poudel2026federated,title={Federated LoRA Adaptation of BiomedCLIP Across Four International Chest X-Ray Cohorts},author={Poudel, Sanjaya and Kunwor, Nirajan and Dhakal, Manish and Jha, Debesh and Gaire, Sunil Kumar},journal={arXiv preprint arXiv:2609.02101},year={2026},}