06 June 2024 : Review article
The Potential of Artificial Intelligence in Prosthodontics: A Comprehensive Review
Ibrahim Saleh Aljulayfi
1ABC*, Ali Hamoud Almatrafi2ABD, Ramzi O. Althubaitiy1ACG, Fahad Alnafisah3BCF, Khalid Alshehri3BCF, Bandar Alzahrani3BCF, Khalid Gufran 4DEG
DOI: 10.12659/MSM.944310
Med Sci Monit 2024; 30:e944310
Table 3 Selected studies for the prosthodontic diagnosis.
| Authors | AI Model | Application | Dataset | Results | Conclusion |
|---|---|---|---|---|---|
| Melike et al (2021) []13 | AI Model Cranio Catch based on a deep CNN method | Diagnostic charting in panoramic radiography | 1084 Panoramic radiograph | Sensitivity values of the crown, implant, and impacted tooth as 0.9674, 0.9615, and 0.9658, respectively | AI model has promising results in detecting dental conditions in OPG radiographs, except for caries and dental calculus |
| Raith et al (2017) []16 | ANN | Dental cusp classification | 129 jaws | Accuracy (93.5%) | Significant improvement has been shown in the post-processing step after the application of greedy algorithm |
| De Angelis et al (2022) []12 | Apox System | Locating dental prosthesis | 120 OPGs | Sensitivity (89%), specificity (98%), AUC (95%). The diagnostic effectiveness had a value of 0.96 | The results reported in this study showed high values of sensitivity and specificity for all the variables analysed |
| Eto et al (2022) []17 | CNN | Image classification of dental restoration | 300 images | Accuracy (95%) recall (0.952), precision (0.957) | Based on the presence or absence of metallic restorations, convulsion neural networks can interpret pictures obtained with an intraoral scanner and categorize molar teeth into 1 of 3 categories (FMC, In, or CNMR) with accuracy reached 95%. |
| Kim et al (2020) []14 | RCNN + Heuristics | Automated numbering and locating of teeth | 303 OPGs | Accuracy 50.9% with IOU (0.5), sensitivity (75.5), specificity (80.4) | Using AI models to analyze dental panoramic radiographs, including implants and bridges, were developed, enabling the possibility of applying AI to orthodontic or implant OPG images of patients |
| Rohrer et al (2022) []15 | CNN | Dental radiograph classification (fillings, crowns, and root canal fillings). AI model trained on smaller tiles to improve performance | 1781 OPGs | F1 score (95%), sensitivity (95%), IOU (93%) | The study revealed that the model performed better with more tiles. Particularly impacted were smaller, less noticeable objects that could be missed in the whole panoramic image. To improve detection performance, this method might be simply modified for use in different applications |
| Takahashi et al (2021) []18 | CNN | Dental arch classification | 1184 dental images | Accuracy (99.6), precision (0.25), recall (1.0), F-measure (0.4) | CNN can be used to categorize and forecast dental arches. This and other AI technologies will enable the creation of systems for creating detachable partial dentures in the future |
| Takahashi et al (2021) []19 | Deep Learning | Detecting restoration and dental prosthesis | 1904 of dental images | Accuracy (99.7%), AUC for the maxilla and mandible were 0.99 and 0.98, respectively | Deep learning has been used to recognize and forecast dental prostheses and restorations with a high accuracy when they are metallic in color, but only moderately effectively when they are tooth-colored |
| Vinayahalingam et al (2021) []4 | Deep Learning based on R-CNN | Radiographic detection of teeth, crowns, filings, root filings, and implants | 2000 OPGs | F1 score (97%), precision (96%), recall (93%) | CNN provides a promising platform for future advancements in automatic chart filling on periapical radiographs |
| Baydar et al (2023) []39 | CNN | Radiographic evaluation of crowns, pulp, restoration, and caries with convolution neural networks | 500 bitewings | Crowns: F1 score (96%), sensitivity (92%), precision (100%) | CNN has the capability to automatically assess bite-wing radiographs |
| Orhan (2023) []40 | Diagnocat AI | Evaluate the accuracy and effectiveness of AI in identification of dental conditions using panoramic radiographs | 4497 teeth | Crowns: sensitivity (90.8%), precision (90.4%) | AI-based decision support systems can serve as a valuable tool in detecting dental conditions, when used with PR for clinical dental applications |
| AI – artificial intelligence; CNN – convolutional neural network; ANN – artificial neural networks; R-CNN – Region-Based Convolutional Neural Network; OPG – orthopantomogram; IOU – intersection over union. | |||||