Abstract
Rheumatoid arthritis (RA) is a common autoimmune disease affecting approximately 1% of the population and causing chronic joint pain and swelling. As progressive inflammation can cause irreversible joint damage, early detection and follow-up are essential. Clinical assessment relies on specialist palpation supported by X-ray, MRI, and ultrasonography, which require equipment and expertise, limiting timely evaluation in telemedicine and specialist-scarce settings. RGB image-based screening provides an accessible, low-cost alternative. However, existing methods are limited by scarce, imbalanced joint-level labels, limited joint coverage and inter-joint context, and insufficient modeling of subtle inflammatory cues. This makes them vulnerable to shifts in skin tone, illumination, and acquisition conditions. To address these limitations, we propose a joint-level RA inflammation detection framework. We also construct RASH, a large-scale synthetic image dataset that spans diverse RA appearances across skin tones and disease severities. We integrate adjacency-based inter-joint context across 11 joints, including the wrist, and guide feature learning toward clinically relevant cues while performing domain alignment only within pathology-focused regions. Experiments on a real patient dataset, including cases that are challenging even for expert visual assessment, demonstrate that our method outperforms baseline approaches and approaches expert-level performance under varying imaging conditions.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/RHEUM_AI_001.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026-sat/supp/RHEUM_AI_001_supp.pdf
Link to Open Review
Open Review Page: https://openreview.net/forum?id=LbATV5KdBK
BibTex
@InProceedings{ChiRin_RASH_MICCAISAT2026,
author = { Chiba, Rintaro AND Kato, Shun AND Kitsunai, Daiki AND Kondo, Yasushi AND Aoki, Yoshimitsu AND Isogawa, Mariko},
title = { { RASH: Rheumatoid Arthritis Synthetic Hand Dataset for Joint-Level Inflammation Estimation } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17271},
month = {pending},
page = {pending}
}