Atlas Construction and Validation Using Cone-Beam Computed Tomography Images of Healthy Adult Human Hands
Anatomical atlases provide standardized spatial references for comparing complex structures. This study develops a high-resolution anatomical atlas of healthy hand bones using cone-beam computed tomography (CBCT), which provides enhanced spatial resolution and lower radiation dose compared to conventional CT. The atlas, constructed from CBCT bilateral scans of 18 healthy adult subjects (35 total hands), serves as a standardized spatial reference for bones in the carpal, metacarpal, and proximal phalangeal regions, where variations in individual scans can complicate analysis. An iterative registration and averaging process, coupled with image preprocessing, enabled the construction of the atlas. The atlas was validated using shape analysis of each bone in the atlas compared to the input images. Validation showed a mean z-score of 0.003 across metrics (volume, surface area, axis lengths, elongation, and flatness), confirming anatomical accuracy. Atlas-based segmentation accuracy, assessed on three independent images, yielded a Dice coefficient of 0.94 and Hausdorff distance of 1.73 mm. Additionally, a subtraction-based method applied to rheumatoid arthritis (RA)-affected joints achieved a mean erosion volume difference of 4.9% compared to manual annotations. This CBCT-based atlas is a valuable reference for segmentation and analysis in bone-related disease research, particularly for tracking RA-induced changes in bone morphology.
Authors
- M. Ethan MacDonald (ORCID: https://orcid.org/0000-0001-5421-3536)
- Abdoljalil Addeh (ORCID: https://orcid.org/0000-0003-2727-4557)
- Sarah L. Manske (ORCID: https://orcid.org/0000-0003-0888-9556)
- Biniyam Mezgebo
- Yousif Al-Khoury (ORCID: https://orcid.org/0009-0005-7197-1845)
Institutions
- University of Calgary (CA)
Publication Details
- Published
- 2025-04-14
- DOI
- https://doi.org/10.1109/isbi60581.2025.10980964
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Canada Foundation for Innovation