An artificial intelligence model for the radiographic diagnosis of osteoarthritis of the temporomandibular joint.

The interpretation of the signs of Temporomandibular joint (TMJ) osteoarthritis on cone-beam computed tomography (CBCT) is highly subjective that hinders the diagnostic process. The objectives of this study were to develop and test the performance of an artificial intelligence (AI) model for the diagnosis of TMJ osteoarthritis from CBCT. A total of 2737 CBCT images from 943 patients were used for the training and validation of the AI model. The model was based on a single convolutional network while object detection was achieved using a single regression model. Two experienced evaluators performed a Diagnostic Criteria for Temporomandibular Disorders (DC/TMD)-based assessment to generate a separate model-testing set of 350 images in which the concluded diagnosis was considered the golden reference. The diagnostic performance of the model was then compared to an experienced oral radiologist. The AI diagnosis showed statistically higher agreement with the golden reference compared to the radiologist. Cohen's kappa showed statistically significant differences in the agreement between the AI and the radiologist with the golden reference for the diagnosis of all signs collectively (P = 0.0079) and for subcortical cysts (P = 0.0214). AI is expected to eliminate the subjectivity associated with the human interpretation and expedite the diagnostic process of TMJ osteoarthritis.

Improving the accuracy of publicly available search engines in recognizing and classifying dental visual assets using convolutional neural networks.

AIM: To assess the accuracy of DigiBrain4, Inc (DB4) Dental Classifier and DB4 Smart Search Engine* in recognizing, categorizing, and classifying dental visual assets as compared with Google Search Engine, one of the largest publicly available search engines and the largest data repository. MATERIALS AND METHODS: Dental visual assets were collected and labeled according to type, category, class, and modifiers. These dental visual assets contained radiographs and clinical images of patients' teeth and occlusion from different angles of view. A modified SqueezeNet architecture was implemented using the TensorFlow r1.10 framework. The model was trained using two NVIDIA Volta graphics processing units (GPUs). A program was built to search Google Images, using Chrome driver (Google web driver) and submit the returned images to the DB4 Dental Classifier and DB4 Smart Search Engine. The categorical accuracy of the DB4 Dental Classifier and DB4 Smart Search Engine in recognizing, categorizing, and classifying dental visual assets was then compared with that of Google Search Engine. RESULTS: The categorical accuracy achieved using the DB4 Smart Search Engine for searching dental visual assets was 0.93, whereas that achieved using Google Search Engine was 0.32. CONCLUSION: The current DB4 Dental Classifier and DB4 Smart Search Engine application and add-on have proved to be accurate in recognizing, categorizing, and classifying dental visual assets. The search engine was able to label images and reject non-relevant results.

Three-dimensional evaluation of the holographic projection in digital dental model superimposition using HoloLens device.

OBJECTIVE: To assess validity and reliability of palatal superimposition of holograms of 3D digital dental models using a customized software, (Ortho Mechanics Sequential Analyzer OMSA), installed on Microsoft HoloLens device as compared to the OMSA application running on a regular computer screen. METHODS: The sample consisted of pre- and post-treatment digital maxillary dental models of 20 orthodontic cases (12.3 ± 1.9 years) treated by rapid maxillary expansion (two turns per day). For each case, the pre- and post-treatment digital models were superimposed using hand gestures for marking the dental models holograms in mixed reality using the Microsoft HoloLens. The same models were then superimposed using the conventional landmark-based method with OMSA software running on a regular computer screen. The same set of dental arch parameters was measured on the superimposed 3D data by the two software versions for comparison. Agreement in the superimposition outcomes among the two superimposition methods was assessed using Dahlberg error (DE), concordance correlation coefficients (CCCs) using two-way ANOVA mixed model for absolute agreement and Bland-Altman analysis. RESULTS: Repeatability was acceptable for all variables based on the high values of CCCs over 0.99 with a lower 95% confidence limit over 0.95 for any variable. The DE ranged from 0.14 mm to 0.36 mm. The absolute error did not exceed 0.5 mm for any variable. CONCLUSION: Using the depth vision capabilities of the Microsoft HoloLens, 3D digital dental models can be reliably superimposed allowing virtual assessment of orthodontic treatment outcomes.

Validity and reliability of three-dimensional palatal superimposition of digital dental models.

OBJECTIVE: To evaluate the validity and reliability of three-dimensional (3D) landmark-based palatal superimposition of digital dental models using Ortho Mechanics Sequential Analyzer (OMSA). METHODS: The sample consisted of pre- and post-treatment digital maxillary dental models of 20 orthodontic cases. For each case, the pre- and post-treatment digital models were superimposed using surface-based methods utilizing 3dMD Vultus and Invivo 5 software as well as a landmark-based method utilizing OMSA. The same set of parameters were measured on the superimposed 3D data by the three softwares for comparison. Agreement in the superimposition outcomes among the three superimposition methods was evaluated with intraclass correlation coefficients (ICCs), Bland-Altman plots, and repeated measures ANOVA. A P value of ≤ 0.05 was considered statistically significant. RESULTS: Repeatability was acceptable for all methods based on the ICCs. Agreement as measured by the ICCs and repeated measures ANOVA was high among the three methods. CONCLUSION: The results indicate that OMSA offers a valid and reliable tool for 3D landmark-based digital dental models superimposition using 3 points marked along the midpalatal raphe as reference.

Reliability of linear and angular dental measurements with the OrthoMechanics Sequential Analyzer.

INTRODUCTION: The aim of this study was to evaluate the reliability of newly developed software in the assessment of orthodontic tooth movement 3 dimensionally. METHODS: The sample consisted of pretreatment and posttreatment computed tomography scans and plaster dental models of 20 orthodontic patients treated with a hyrax palatal expander as a part of their comprehensive orthodontic treatment. Dental-arch measurements, including arch widths, tooth inclinations, and angulation parameters, were measured on the scans using InvivoDental 3D imaging software (version 5.1; Motionview, Hixson, Tenn). The plaster dental models were laser scanned and superimposed, and measurements were obtained digitally using the new software. Agreement between the digital models and the computed tomography measurements was evaluated with intraclass correlation coefficients, paired t tests, and Bland-Altman plots. A P value of ≤0.05 was considered statistically significant. RESULTS: High agreement, a nonsignificant paired t test, and no indication of agreement discrepancies were observed for most of the measured parameters. CONCLUSIONS: The results confirmed that the new software program offers a reliable tool for dental-arch measurements obtained from 3-dimensional laser-scanned models.