Root canal segmentation from cone-beam computed tomography guided by micro-computed tomography based on deep learning

Abstract Background Accurate root canal segmentation from cone-beam computed tomography (CBCT) is essential for endodontic diagnosis and treatment planning. This study aims to explore the feasibility of using deep learning (DL) models, trained on CBCT images of extracted teeth guided by micro-computed tomography (µCT), for clinical CBCT image segmentation. Methods A dataset of 56 extracted teeth with diverse root canal complexities was constructed, combining CBCT and µCT scans. Ground truth annotations were derived from µCT-based masks and registered to CBCT images. DL models based on U²-Net architecture were trained and evaluated for tooth and root canal segmentation, comparing µCT-guided and manual-label-based approaches. The effects of field-of-view (FOV) size and interpolation algorithms on segmentation performance were investigated. The trained models were applied to clinical CBCT images, achieving rapid and accurate root canal segmentation validated by endodontic specialists. Results The µCT-guided AI segmentation method outperformed the manual-label-based approach. Combining a limited FOV with an interpolation algorithm demonstrated notable advantages in capturing intricate details. In segmenting root canal from patient CBCT images, 94% of single rooted teeth and 100% of molars, were rated as “excellent” or “good”. Conclusions Results demonstrated the potential of µCT-guided DL models for enhancing root canal segmentation in clinical practice, offering a promising tool for digital dentistry. Clinical trial number: not applicable