Beyond Segmentation: Coronary Artery Calcium Scoring with Frozen Self-Supervised CT Features
Journal of Imaging · 16.11.2026 · 12 (11) · 579
DOI: 10.3390/jimaging12110579 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Imaging, online erschienen am 16.11.2026. Verfasst von Mahmut S. Gokmen und 11 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University of Kentucky, National Center for Advancing Translational Sciences.
Abstract (Originalsprache)
Coronary artery calcium (CAC) scoring on computed tomography (CT) acquired for other indications could provide calcium-burden information without a dedicated scoring examination. We investigate direct, segmentation-free Agatston score prediction from frozen CT patch representations using only volume-level supervision for the scoring head. A lightweight additive head maps attention-selected features to non-negative contributions, combines them within slices, and sums slice scores across each volume. In developmental evaluation, applying the nonlinear mapping before aggregation improved linearly weighted κ by 0.27–0.38 over a pooled multilayer perceptron across six encoders. Using spatially adaptive label-guided temperature (SALT) features, held-out κ was 0.930 on gated CT and 0.844 on contrast-enhanced coronary CT angiography (CCTA), with separately fitted heads. On external non-gated CT, a head fitted on 165 patients from the same cohort achieved κ=0.727 on 40 held-out patients; patient-level five-fold cross-validation over all 205 patients yielded κ=0.639, indicating preliminary local-adaptation performance. Unchanged-head external gated transfer achieved κ=0.854 (n=443). Gated intraclass correlation coefficient ICC(A,1) on log(1+score) was 0.984. Study-level CAC scores served as CCTA reference targets; non-contrast images were not model inputs. SALT uses gated calcium masks during pretraining, not downstream scoring. Its highest-attention patch overlapped calcium in 89.5% of calcium-bearing slices, but localization rankings did not directly track scoring rankings. These results support direct, spatially decomposable CAC prediction from reusable features; broader validation is needed before clinical use.
Förderung
- University of Kentucky
- National Center for Advancing Translational Sciences
Zitieren
APA
Gokmen, M. S., Damron, E., Leach, C., Haque, M., Parker, C. S., Hobbs, S., Oates, M. E., Sorrell, V. L., Zhang, J., Leung, S. W., Seales, W. B., & Bumgardner, V. K. C. (2026). Beyond Segmentation: Coronary Artery Calcium Scoring with Frozen Self-Supervised CT Features. Journal of Imaging, 12(11), 579. https://doi.org/10.3390/jimaging12110579
Vancouver
Gokmen MS, Damron E, Leach C, Haque M, Parker CS, Hobbs S, et al. Beyond Segmentation: Coronary Artery Calcium Scoring with Frozen Self-Supervised CT Features. Journal of Imaging. 2026;12(11):579. doi:10.3390/jimaging12110579
BibTeX
@article{Gokmen2026,
author = {Gokmen, Mahmut S. and Damron, Evan and Leach, Caroline and Haque, Moneera and Parker, C. Seth and Hobbs, Stephen and Oates, M. Elizabeth and Sorrell, Vincent L. and Zhang, Jie and Leung, Steve W. and Seales, W. Brent and Bumgardner, V. K. Cody},
title = {Beyond Segmentation: Coronary Artery Calcium Scoring with Frozen Self-Supervised CT Features},
journal = {Journal of Imaging},
year = {2026},
volume = {12},
number = {11},
pages = {579},
doi = {10.3390/jimaging12110579}
}
EUVIMED