Clinical Operating-Point Evaluation of a CLAHE-Enhanced Tiled YOLO۱۱s Framework for Proximal Caries Detection on Bitewing Radiographs

Interproximal caries are difficult to identify on bitewing radiographs because lesions are small, low contrast, and located at contact surfaces where anatomical overlap may obscure early changes. Rather than proposing a new neural-network architecture, this study evaluated a deployment-oriented framework for image-level proximal caries detection using contrast-limited adaptive histogram equalization (CLAHE), overlapping tiles, and YOLO۱۱s. The dataset contained ۱,۰۸۴ bitewing radiographs from ۳۰۶ patients, including ۵۸۵ caries-positive and ۴۹۹ normal images with ۱,۹۱۵ expert-annotated lesions. Preliminary comparison of full-image YOLO۱۱s, clean tiled YOLO۱۱s, and CLAHE-enhanced tiled YOLO۱۱s supported selection of the CLAHE-tiled configuration for final evaluation. Five-fold patient-wise cross-validation with an inner patient-wise validation split ensured that every radiograph was evaluated once in a held-out test fold. Training retained all eligible negative tiles, whereas validation and test radiographs were processed exhaustively without annotation-guided region selection. Inspired by clinical operating-point evaluation in chest radiography, three prespecified validation-derived modes represented sensitivity-prioritized screening, balanced classification, and specificity-prioritized false-positive control. The sensitivity-prioritized mode achieved ۹۱.۹۴% ± ۴.۸۸% sensitivity and ۴۳.۱۸% ± ۲۱.۵۹% specificity; the balanced mode achieved ۸۷.۹۸% ± ۴.۹۳% sensitivity and ۵۸.۱۱% ± ۱۹.۷۶% specificity, with the highest F۱-score of ۷۸.۷۰% ± ۴.۵۹%; and the specificity-prioritized mode achieved ۶۹.۶۲% ± ۱۷.۵۶% sensitivity and ۷۷.۷۸% ± ۱۰.۲۲% specificity. The principal contribution is methodological and evaluation-oriented, combining patient-wise exhaustive inference, validation-only operating-point selection, and explicit assessment of sensitivity-specificity trade-offs. The patient-wise exhaustive evaluation provides internal evidence for this framework, while between-fold variability and the single-center design indicate that external validation and prospective workflow evaluation are required before clinical deployment.

کلیدواژه ها:

bitewing radiography ، proximal caries ، YOLO۱۱s ، CLAHE ، clinical operating points ، exhaustive evaluation ، diagnostic bias