Related Experiment Video
Updated: Oct 2, 2025
A Postoperative Evaluation Guideline for Computer-Assisted Reconstruction of the Mandible
Published on: January 28, 2020
Detection and classification of mandibular fracture on CT scan using deep convolutional neural network
Xuebing Wang1, Zineng Xu2, Yanhang Tong1
1Department of Oral and Maxillofacial SurgeryNational Engineering Laboratory for Digital and Material Technology of Stomatology; Beijing Key Laboratory of Digital StomatologyNational Clinical Research Center for Oral Diseases, Peking University School and Hospital of Stomatology, No 22 Zhongguancun South Road, Beijing, 100081, People's Republic of China.
Convolutional neural networks (CNNs) accurately detect and classify mandibular fractures on CT scans. This AI approach offers reliable diagnostic support, especially in underserved medical areas.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in radiology
Background:
- Mandibular fractures are common injuries requiring accurate diagnosis.
- Computed tomography (CT) is the standard imaging modality.
- Manual interpretation of CT scans can be time-consuming and subjective.
Purpose of the Study:
- To assess the accuracy and reliability of convolutional neural networks (CNNs) for detecting and classifying mandibular fractures.
- To evaluate the performance of a CNN-based algorithm against expert annotations.
Main Methods:
- A dataset of 686 patient CT scans with mandibular fractures was used.
- Three experienced maxillofacial surgeons provided ground truth annotations.
- Two CNN models (U-Net and ResNet) were trained and validated.
- Performance metrics included DICE, accuracy, sensitivity, specificity, and AUC.
Main Results:
- The U-Net model achieved a mandible segmentation DICE of 0.943.
- CNNs demonstrated accuracies above 90% across nine mandibular subregions.
- The mean area under the ROC curve (AUC) was 0.956, indicating high diagnostic performance.
Conclusions:
- CNNs provide reliable and accurate detection and classification of mandibular fractures on CT.
- This AI algorithm can enhance diagnostic efficiency and extend expertise to areas with limited medical resources.
