Back to search

University of British Columbia

Pre-implant tracing of mandibular canals on CBCT images using artificial intelligence

Abstract

dc:description

Background: Accurate localization of mandibular canals (MC) is important in Dental Implantology to avoid injuries of mandibular nerve inside the canal. Artificial intelligence (AI) has impacts on improving the quality of delivering Implantology. The aim of this study was to compare the accuracy of an AI system with conventional methods in detecting MC during pre-implant planning using cone-beam computed tomography (CBCT) images. Methods: A pilot study of six CBCT scans of a dentate human mandible was initially conducted to explore measurement agreements of manual tracing (coDiagnostiX), AI method (Diagnocat) and histology sections as gold standard. Measurements were conducted in 2nd premolars and 1st and 2nd molars areas bilaterally and coded as LM2, LM1, LP2, RP2, RM1, RM2. Subsequently, a retrospective study of 47 patients’ CBCT scans of their mandibles was analysed for MC localization by 2 examiners using coDiagnostiX, and Diagnocat. Measurements were taken from the midpoint of the superior aspect of the buccal and lingual cortex to the superior aspect of MC in the regions of interest (ROIs). The data was calibrated for intra-examiner and inter-examiner reliability. One-way ANOVA test was used to analyse three sets of measurements in pilot studies and independent samples t-tests were used in retrospective studies. Results: 110 CBCTs were screened, 57 CBCTs were excluded due to different ROIs, 2 CBCTs were excluded in coDiagnostiX due to old Digital Imaging and Communications in Medicine (DICOM) files and 4 CBCTs were excluded in Diagnocat due to metal artifact or extended ROI outside mandible area. Twelve CBCTs in AI software showed incidental findings including thickening of maxillary sinus membrane or dense bone islands in the mandible. No statistically significant difference found in the pilot study among the measurements of three groups except for RM2 (p<0.05) likely due to complex anatomy of the area and increased bone thickness. When sample size increased in the retrospective study, no statistically significant difference found between AI and coDiagnostiX in edentulous and dentate groups or both groups combined using independent samples t-test. Conclusion: AI software used in this study is as accurate and predictable as manual tracing software for MC detection.

Degree

thesis:*
Name thesis:degree_name
Master of Science - MSc
Level thesis:degree_level
master's
Discipline thesis:degree_discipline
Craniofacial Science
Grantor dc:publisher
University of British Columbia
Year dc:date
2023

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Dadoush, Maher
Contributors dc:contributor
  • MacDonald, David, 1955-

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/85371

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Dadoush, Maher. Pre-implant tracing of mandibular canals on CBCT images using artificial intelligence. master's thesis, University of British Columbia, 2023. http://hdl.handle.net/2429/85371