Abstract
Juvenile idiopathic arthritis (JIA) can affect the temporo- mandibular joint (TMJ), leading to condylar alterations and facial de- formities. Early diagnosis is crucial to limit disease-related damage, yet clinical assessment alone is often insufficient. Therefore, the evaluation of mandibular condylar asymmetry through the identification of specific anatomical landmarks becomes an important tool for the assessment of TMJ involvement. Orthopantomography (OPG) is a widely used, low- dose imaging modality for this purpose; however, manual landmark an- notation remains time-consuming and highly operator-dependent. While previous studies have explored automatic landmark detection in healthy subjects, we are the first to investigate it in a challenging pathological context like JIA, where normal anatomy is significantly altered and inter- operator variability is prominent. In this work, we present a pipeline for automatic condylar asymmetry assessment on OPG images. The pipeline combines a web-based annotation tool, a deep learning model localizing six clinically relevant landmarks, and a systematic evaluation of landmark-derived asymmetry indexes for normal-versus-pathological classification. Our model achieved a global mean radial error of 1.58 ± 2.31 mm, with 94.06% of landmarks detected within 4 mm. Measurement agreement analysis showed satisfactory consistency between predicted and manual landmarks, although highly modified pathological morpholo- gies remained challenging. The asymmetry-index evaluation confirmed the discriminative relevance of condylar measures, including the Habets index, while highlighting a trade-off between classification performance and robustness to landmark prediction errors. These findings support the feasibility of automatic OPG-based condylar asymmetry assessment in challenging clinical scenarios.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ODIN_005.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=PjqjF8e1QU
BibTex
@InProceedings{SigMic_AIBased_MICCAISAT2026,
author = { Signori, Michele AND Savardi, Mattia AND Cattadori, Martina AND Tonni, Ingrid AND Signoroni, Alberto},
title = { { AI-Based Landmark Detection for Condylar Asymmetry Assessment in Patients with Juvenile Idiopathic Arthritis } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17270},
month = {pending},
page = {pending}
}
