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 4 Selected studies for the prediction of prosthodontic treatment outcome.
| Authors | AI Model | Application | Datasets | Results | Conclusions |
|---|---|---|---|---|---|
| Chen et al (2020) []28 | Fuzzy decision | Dental color matching | 26 photos | Accuracy (99.78%) | The fuzzy decision approach incorporates the HSV color model, PSNR(H), PSNR(S), and SSIM information, which are employed for the first time in a tooth color-matching study. The results suggest that the proposed strategy outperforms previous approaches |
| Cheng et al (2015) []20 | BP neural network | Predicting facial deformation following complete denture | 48 patients | Average error rate (22.49%) | Deformation and postoperative models are superimposed. The results of the experiments revealed that BP neural networks can predict face soft tissue deformation rapidly and accurately |
| Deniz S et al (2019) []22 | ANN | Prediction of surface roughness and microhardness of various denture teeth | 10 maxillary first and second molars | Mean square error for microhardness (7.6596) and surface roughness | ANN can predict the properties with a relatively low prediction error. Surface roughness values rose significantly with time in all beverages, and micro filler composite resin denture teeth had the best mechanical and physical qualities |
| Kim et al (2018) []26 | SVM | Dental shade determination | 26 shade tabs | Accuracy (96.9%) | The SVM model could be an ideal method for quantitatively measuring tooth color with high accuracy |
| Lee et al (2022) []25 | Decision tree | Tooth prognosis decision | 2359 teeth from 94 cases | Accuracy (84%), standard deviation (93%) | the use of AI in the decision-making process for tooth prognosis in relation to the treatment plan |
| Li et al (2015) []5 | BPNN | Predicting the quality of color matching in restorative dentistry | 119 porcelain specimens | Error rate (0.042) | In practical research, the GA+BP can aid in achieving the Computer Color Matching prediction target. As a result, it has a high practical application value and serves as a guide in CCM. With the complete introduction of computer science into the medical sphere, stomatological hospitals will be able to give greater services to patients in the future |
| Li et al (2022) []23 | ET | Predicting the flexural strength of CAD/CAM RCBs | 16 fillers and monomers | Accuracy (99.9%) | AI models developed accurate predictions of the flexural strength of CAD/CAM RCBs and identified the effective components that affected flexural strength |
| Justiawan et al (2019) []29 | KNN | Dental image classification | 16 shade guides | Accuracy (97.5%) | When compared to the other alternatives, the KNN algorithm in RGB characteristics delivers outstanding performance |
| Ueki et al (2020) []27 | NN | True color prediction | 62 dental photos | Accuracy (68%) | The study validated the neural network model’s efficacy. The image database that can be utilized for training is currently relatively minimal. However, we discovered that by taking a large number of training data points from a single image and utilizing them to train a neural network, we could get relatively accurate color candidates |
| Wei et al (2018) []8 | BPNN | Color matching | 43 PFM specimens | Color differences (0.95–1.26) | Computer Color Matching system produced higher accuracy in color reproduction within the given color space than the gold standard approach |
| Yamaguchi et al (2019) []24 | CNN | Predict the debonding of CR crowns | 8640 dental images | Accuracy (98.5%), precision (97%), recall (100%) | AI technology, specifically deep learning with a CNN approach, performed outstandingly in forecasting the debonding probability of a CAD/CAM CR crown using 3D stereolithography models of a die scanned from patients |
| Yuan F et al (2017) []21 | BPNN | Predicting the aesthetic reconstruction of edentulous patients after wearing complete denture | 10 patients | Error rate (0.480–1.090) | To a certain extent, this technology can achieve virtual prediction of soft tissue appearance in the lower area of the face after wearing complete dentures |
| Yang J et al (2023) []41 | ML | predict the final colors of CAD-CAM ceramics and determine their required minimum thicknesses to cover different clinical backgrounds | 120 ceramic specimens | Prediction error (ΔE <2.6) | Machine learning provided a promising tool for automate decision support in material selection with minimal thickness over various clinical application |
| AI – artificial intelligence; BP – backpropagation; ANN – artificial neural networks; SVM – support vector machine; BPNN – the network architecture of backpropagation neural network; ET – extra tree; CAD/CAM – computer-aided design/computer-aided manufacturing; KNN – K-Nearest Neighbors; NN – neural network; CNN – convolutional neural network; RGB – red, green, blue; HSV – hue, saturation, and value. | |||||