Accurate segmentation of the Inferior Alveolar Canal (IAC) in Cone-Beam Computed Tomography (CBCT) scans is a clinically important yet technically demanding task. Despite growing interest in automated approaches, the field has long lacked publicly available datasets and standardized benchmarks, making systematic comparison between methods difficult.
To address this gap, the ToothFairy challenge was organized as part of the MICCAI 2023 conference. A dedicated public dataset of 443 CBCT scans was released, with voxel-level IAC annotations available for 153 of them — the largest resource of its kind to date. Participants were challenged to develop algorithms capable of accurately identifying the IAC from both 2D and 3D annotated data.
This paper documents the challenge in detail and evaluates the most promising submitted solutions, offering the first comprehensive comparison of IAC segmentation methods on a shared benchmark. Beyond summarizing the current state of the art, the authors identify key open challenges and outline directions for future research. To support reproducibility and continued development, an open-source repository collecting the best-performing implementations has also been made available.
Published by an international team of researchers. Among the co-authors are Sano scientists: Tomasz Szczepański, Michal K. Grzeszczyk and Przemyslaw Korzeniowski.
Read the article