Accurate mandibular canal segmentation of dental CBCT using a two-stage 3D-UNet based segmentation framework

• Published in: BMC oral health Vol. 23; no. 1; pp. 551 - 7
• Main Authors: Lin, Xi, Xin, Weini, Huang, Jingna, Jing, Yang, Liu, Pengfei, Han, Jingdan, Ji, Jie
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 10.08.2023; BioMed Central; BMC
• Subjects: Analysis; Artificial intelligence; Computed tomography; Cone-beam computerized tomography; CT imaging; Cysts; Deep learning; Dental radiology; Energy consumption; Entropy; Ethics; Inferior alveolar nerve; Machine learning; Mandibular canal; Methods; Neural networks; Orthodontics; Osteotomy; Patient outcomes; Segmentation; Software; Surgery; Teeth; Transplants & implants; Values; China; Dental radiology; Cone-beam computerized tomography; Artificial intelligence; Inferior alveolar nerve
• ISSN: 1472-6831; 1472-6831
• DOI: 10.1186/s12903-023-03279-2

The objective of this study is to develop a deep learning (DL) model for fast and accurate mandibular canal (MC) segmentation on cone beam computed tomography (CBCT). A total of 220 CBCT scans from dentate subjects needing oral surgery were used in this study. The segmentation ground truth is annotated and reviewed by two senior dentists. All patients were randomly splitted into a training dataset (n = 132), a validation dataset (n = 44) and a test dataset (n = 44). We proposed a two-stage 3D-UNet based segmentation framework for automated MC segmentation on CBCT. The Dice Similarity Coefficient (DSC) and 95% Hausdorff Distance (95% HD) were used as the evaluation metrics for the segmentation model. The two-stage 3D-UNet model successfully segmented the MC on CBCT images. In the test dataset, the mean DSC was 0.875 ± 0.045 and the mean 95% HD was 0.442 ± 0.379. This automatic DL method might aid in the detection of MC and assist dental practitioners to set up treatment plans for oral surgery evolved MC.