DIR-YOLO: Segmentation of Alveolar Bone and Mandibular Canal in CBCT Images for Dental Implant Recommendation
Author(s) -
Mohammad Farid Naufal,
Chastine Fatichah,
Eha Renwi Astuti,
Ramadhan Hardani Putra
Publication year - 2025
Publication title -
ieee access
Language(s) - English
Resource type - Magazines
SCImago Journal Rank - 0.587
H-Index - 127
eISSN - 2169-3536
DOI - 10.1109/access.2025.3621623
Subject(s) - aerospace , bioengineering , communication, networking and broadcast technologies , components, circuits, devices and systems , computing and processing , engineered materials, dielectrics and plasmas , engineering profession , fields, waves and electromagnetics , general topics for engineers , geoscience , nuclear engineering , photonics and electrooptics , power, energy and industry applications , robotics and control systems , signal processing and analysis , transportation
In dental implant planning, segmentation of the alveolar bone (AB) and mandibular canal (MC) is essential to identify a safe area for dental implant placement in the edentulous molar region. This anatomical information serves as the basis for radiologists to determine the most appropriate dental implant dimension, including its diameter and height. Cone-Beam Computed Tomography (CBCT) is a 3D imaging modality that provides high-resolution cross-sectional images of the jaw. It enables detailed visualization of AB and MC. The dental radiologists manually identify the AB and MC to determine the suitable dental implant dimension. This process is time-consuming and its accuracy depends on the dental radiologist’s skill and experience. This study proposes an automated method for segmenting AB and MC from CBCT images to support dental implant dimension recommendation for molar teeth. This study introduces DIR-YOLO (Dental Implant Recommendation-YOLO), an efficient AB and MC segmentation model based on a modified YOLOv8 architecture. The proposed method enhances YOLOv8 by replacing the original C2f module with C3Ghost, replacing the standard convolution with Ghost Convolution to reduce computational complexity, and simplifying the segmentation head to use only two-scale feature maps. It focuses on low and high resolutions for optimized AB and MC segmentation performance. DIR-YOLO achieved a mean Dice Similarity Coefficient (mDSC) of 92.12% and mean Hausdorff Distance (mHD) of 0.96 mm, indicating high AB and MC segmentation performance. Furthermore, Weighted Cohen’s Kappa and the Kruskal-Wallis test demonstrate a high level of statistical agreement between the proposed method and the dental radiologists and indicating no significant differences in the recommended dental implant dimensions. These results indicate that DIR-YOLO can assist dental radiologists by automating AB and MC segmentation for supporting accurate dental implant dimension recommendation for molar teeth.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom