Advantages of deep learning with convolutional neural network in detecting disc displacement of the temporomandibular joint in magnetic resonance imaging

Abstract

This study investigated the usefulness of deep learning-based automatic detection of anterior disc displacement (ADD) from magnetic resonance imaging (MRI) of patients with temporomandibular joint disorder (TMD). Sagittal MRI images of 2520 TMJs were collected from 861 men and 399 women (average age 37.33 ± 18.83 years). A deep learning algorithm with a convolutional neural network was developed. Data augmentation and the Adam optimizer were applied to reduce the risk of overfitting the deep-learning model. The prediction performances were compared between the models and human experts based on areas under the curve (AUCs). The fine-tuning model showed excellent prediction performance (AUC = 0.8775) and acceptable accuracy (approximately 77%). Comparing the AUC values of the from-scratch (0.8269) and freeze models (0.5858) showed lower performances of the other models compared to the fine-tuning model. In Grad-CAM visualizations, the fine-tuning scheme focused more on the TMJ disc when judging ADD, and the sparsity was higher than that of the from-scratch scheme (84.69% vs. 55.61%, p < 0.05). The three fine-tuned ensemble models using different data augmentation techniques showed a prediction accuracy of 83%. Moreover, the AUC values of ADD were higher when patients with TMD were divided by age (0.8549–0.9275) and sex (male: 0.8483, female: 0.9276). While the accuracy of the ensemble model was higher than that of human experts, the difference was not significant (p = 0.1987–0.0671). Learning from pre-trained weights allowed the fine-tuning model to outperform the from-scratch model. Another benefit of the fine-tuning model for diagnosing ADD of TMJ in Grad-CAM analysis was the deactivation of unwanted gradient values to provide clearer visualizations compared to the from-scratch model. The Grad-CAM visualizations also agreed with the model learned through important features in the joint disc area. The accuracy was further improved by an ensemble of three fine-tuning models using diversified data. The main benefits of this model were the higher specificity compared to human experts, which may be useful for preventing true negative cases, and the maintenance of its prediction accuracy across sexes and ages, suggesting a generalized prediction.

Assisting in Diagnosis of Temporomandibular Disorders: A Deep Learning Approach

“…Moreover, the high sensitivity and specificity mean that the proposed model could accurately distinguish the TMD patients and non-TMD patients. We could also compare the results with existing TMD-related research [28], [29], [34]. We could find the ANN model could achieve the accuracy rate of 92.31%, which was much higher than the CNN model [28] and XGBoost+LightGBM [29] with the accuracy rates of 76.92% and 82.3%, respectively.…”

“…We could also compare the results with existing TMD-related research [28], [29], [34]. We could find the ANN model could achieve the accuracy rate of 92.31%, which was much higher than the CNN model [28] and XGBoost+LightGBM [29] with the accuracy rates of 76.92% and 82.3%, respectively. The ANN also had lower expense as the variables in the input were easily to be obtained.…”

“…The data in the testing group were new for the constructed ANN model. For the pooled accuracy, the sensitivity and specificity were calculated with the predicting results of testing group obtained by Youden Index [28]. With the data in the testing group, the prediction results of four values including true positive, true negative, false positive, and false negative are 12, 8, 1, and 1, respectively.…”

“…In the field of TMD, deep/machine learning techniques have been considered to improve the diagnosis systems. In 2022, Lee et al [28] utilize the Convolutional Neural Networks (CNN) to detect TMD with the magnetic resonance imaging (MRI), which showed qualified accuracy rate. In 2021, Luciano et al [14] reported results of a comparative analysis of the High Frequency Value (HFV) method with other machine learning models as a validation of the dataset creation process and best classification results obtained by using a combination of classifiers for orofacial pain, headache and TMD diagnosis system.…”

“…From the above literature review, we can find the proposed CNN-based diagnosis system can achieve qualified accuracy rate to detect TMD [28], but requires the MRI results as input, which poses high expense to the patients and cannot be adopted for many hospital/clinics with no MRI machines. The existing ANN-based research focuses on the analysis of potential effects of environmental factors on the possibility of TMD, instead of the diagnosis system development.…”