Deep learning-based evaluation of the relationship between mandibular third molar and mandibular canal on CBCT.

OBJECTIVES: The objective of our study was to develop and validate a deep learning approach based on convolutional neural networks (CNNs) for automatic detection of the mandibular third molar (M3) and the mandibular canal (MC) and evaluation of the relationship between them on CBCT. MATERIALS AND METHODS: A dataset of 254 CBCT scans with annotations by radiologists was used for the training, the validation, and the test. The proposed approach consisted of two modules: (1) detection and pixel-wise segmentation of M3 and MC based on U-Nets; (2) M3-MC relation classification based on ResNet-34. The performances were evaluated with the test set. The classification performance of our approach was compared with two residents in oral and maxillofacial radiology. RESULTS: For segmentation performance, the M3 had a mean Dice similarity coefficient (mDSC) of 0.9730 and a mean intersection over union (mIoU) of 0.9606; the MC had a mDSC of 0.9248 and a mIoU of 0.9003. The classification models achieved a mean sensitivity of 90.2%, a mean specificity of 95.0%, and a mean accuracy of 93.3%, which was on par with the residents. CONCLUSIONS: Our approach based on CNNs demonstrated an encouraging performance for the automatic detection and evaluation of the M3 and MC on CBCT. Clinical relevance An automated approach based on CNNs for detection and evaluation of M3 and MC on CBCT has been established, which can be utilized to improve diagnostic efficiency and facilitate the precision diagnosis and treatment of M3.

Detection and classification of mandibular fracture on CT scan using deep convolutional neural network.

OBJECTIVES: This study aimed to evaluate the accuracy and reliability of convolutional neural networks (CNNs) for the detection and classification of mandibular fracture on spiral computed tomography (CT). MATERIALS AND METHODS: Between January 2013 and July 2020, 686 patients with mandibular fractures who underwent CT scan were classified and annotated by three experienced maxillofacial surgeons serving as the ground truth. An algorithm including two convolutional neural networks (U-Net and ResNet) was trained, validated, and tested using 222, 56, and 408 CT scans, respectively. The diagnostic performance of the algorithm was compared with the ground truth and evaluated by DICE, accuracy, sensitivity, specificity, and area under the ROC curve (AUC). RESULTS: One thousand five hundred six mandibular fractures in nine subregions of 686 patients were diagnosed. The DICE of mandible segmentation using U-Net was 0.943. The accuracies of nine subregions were all above 90%, with a mean AUC of 0.956. CONCLUSIONS: CNNs showed comparable reliability and accuracy in detecting and classifying mandibular fractures on CT. CLINICAL RELEVANCE: The algorithm for automatic detection and classification of mandibular fractures will help improve diagnostic efficiency and provide expertise to areas with lower medical levels.

Towards clinically applicable automated mandibular canal segmentation on CBCT.

OBJECTIVES: To develop a deep learning-based system for precise, robust, and fully automated segmentation of the mandibular canal on cone beam computed tomography (CBCT) images. METHODS: The system was developed on 536 CBCT scans (training set: 376, validation set: 80, testing set: 80) from one center and validated on an external dataset of 89 CBCT scans from 3 centers. Each scan was annotated using a multi-stage annotation method and refined by oral and maxillofacial radiologists. We proposed a three-step strategy for the mandibular canal segmentation: extraction of the region of interest based on 2D U-Net, global segmentation of the mandibular canal, and segmentation refinement based on 3D U-Net. RESULTS: The system consistently achieved accurate mandibular canal segmentation in the internal set (Dice similarity coefficient [DSC], 0.952; intersection over union [IoU], 0.912; average symmetric surface distance [ASSD], 0.046 mm; 95% Hausdorff distance [HD95], 0.325 mm) and the external set (DSC, 0.960; IoU, 0.924; ASSD, 0.040 mm; HD95, 0.288 mm). CONCLUSIONS: These results demonstrated the potential clinical application of this AI system in facilitating clinical workflows related to mandibular canal localization. CLINICAL SIGNIFICANCE: Accurate delineation of the mandibular canal on CBCT images is critical for implant placement, mandibular third molar extraction, and orthognathic surgery. This AI system enables accurate segmentation across different models, which could contribute to more efficient and precise dental automation systems.

Robust automated teeth identification from dental radiographs using deep learning.

OBJECTIVES: This study developed and validated a deep learning-based method to automatically segment and number teeth in panoramic radiographs across primary, mixed, and permanent dentitions. METHODS: A total of 6,046 panoramic radiographs were collected and annotated. The dataset encompassed primary, mixed and permanent dentitions and dental abnormalities such as tooth number anomalies, dental diseases, dental prostheses, and orthodontic appliances. A deep learning-based algorithm consisting of a U-Net-based region of interest extraction model, a Hybrid Task Cascade-based teeth segmentation and numbering model, and a post-processing procedure was trained on 4,232 images, validated on 605 images, and tested on 1,209 images. Precision, recall and Intersection-over-Union (IoU) were used to evaluate its performance. RESULTS: The deep learning-based teeth identification algorithm achieved good performance on panoramic radiographs, with precision and recall for teeth segmentation and numbering exceeding 97%, and the IoU between predictions and ground truths reaching 92%. It generalized well across all three dentition stages and complex real-world cases. CONCLUSIONS: By utilizing a two-stage training framework with a large-scale heterogeneous dataset, the automatic teeth identification algorithm achieved a performance level comparable to that of dental experts. CLINICAL SIGNIFICANCE: Deep learning can be leveraged to aid clinical interpretation of panoramic radiographs across primary, mixed, and permanent dentitions, even in the presence of real-world complexities. This robust teeth identification algorithm could contribute to the future development of more advanced, diagnosis- or treatment-oriented dental automation systems.