Development of an artificial intelligence-based algorithm to classify images acquired with an intraoral scanner of individual molar teeth into three categories

Background Forensic dentistry identifies deceased individuals by comparing postmortem dental charts, oral-cavity pictures and dental X-ray images with antemortem records. However, conventional forensic dentistry methods are time-consuming and thus unable to rapidly identify large numbers of victims following a large-scale disaster. Objective Our goal is to automate the dental filing process by using intraoral scanner images. In this study, we generated and evaluated an artificial intelligence-based algorithm that classified images of individual molar teeth into three categories: (1) full metallic crown (FMC); (2) partial metallic restoration (In); or (3) sound tooth, carious tooth or non-metallic restoration (CNMR). Methods A pre-trained model was created using oral-cavity pictures from patients. Then, the algorithm was generated through transfer learning and training with images acquired from cadavers by intraoral scanning. Cross-validation was performed to reduce bias. The ability of the model to classify molar teeth into the three categories (FMC, In or CNMR) was evaluated using four criteria: precision, recall, F-measure and overall accuracy. Results The average value (variance) was 0.952 (0.000140) for recall, 0.957 (0.0000614) for precision, 0.952 (0.000145) for F-measure, and 0.952 (0.000142) for overall accuracy when the algorithm was used to classify images of molar teeth acquired from cadavers by intraoral scanning. Conclusion We have created an artificial intelligence-based algorithm that analyzes images acquired with an intraoral scanner and classifies molar teeth into one of three types (FMC, In or CNMR) based on the presence/absence of metallic restorations. Furthermore, the accuracy of the algorithm reached about 95%. This algorithm was constructed as a first step toward the development of an automated system that generates dental charts from images acquired by an intraoral scanner. The availability of such a system would greatly increase the efficiency of personal identification in the event of a major disaster.

Accuracy of photogrammetry, intraoral scanning, and conventional impression techniques for complete-arch implant rehabilitation: an in vitro comparative study

Background To compare the accuracy of photogrammetry, intraoral scanning and conventional impression techniques for complete-arch implant rehabilitation. Methods A master cast containing 6 implant abutment replicas was fabricated. Group PG: digital impressions were taken 10 times using a photogrammetry system; Group IOS: intraoral scanning was performed to fabricate 10 digital impressions; Group CNV: splinted open-tray impression technique was used to fabricate 10 definitive casts. The master cast and conventional definitive casts were digitized with a laboratory reference scanner. For all STL files obtained, scan bodies were converted to implant abutment replicas using a digital library. The accuracy of a digitizer was defined by 2 main parameters, trueness and precision. "Trueness" was used to describe the deviation between test files and reference file, and "precision" was used to describe the closeness between test files. Then, the trueness and precision of three impression techniques were evaluated and statistically compared (α = 0.05). Results The median trueness was 24.45, 43.45 and 28.70 μm for group PG, IOS and CNV; Group PG gave more accurate trueness than group IOS (P < 0.001) and group CNV (P = 0.033), group CNV showed more accurate trueness than group IOS (P = 0.033). The median precision was 2.00, 36.00 and 29.40 μm for group PG, IOS and CNV; Group PG gave more accurate precision than group IOS (P < 0.001) and group CNV (P < 0.001), group CNV showed more accurate precision than IOS (P = 0.002). Conclusions For complete-arch implant rehabilitation, the photogrammetry system showed the best accuracy of all the impression techniques evaluated, followed by the conventional impression technique, and the intraoral scanner provided the least accuracy.

Accuracy of printed models obtained from intraoral scanning

Objectives: To compare the accuracy of two methods for the manufacturing of physical models: I) intraoral scanning and resin-printed models; and II) addition silicone impression and gypsum model. Materials and methods: A dental manikin was used as the master model and compared with five gypsum models (g1) and five resin printed models (g2) by analyzing linear measurements at four sites (M1, M2, M3, and M4) using an image measuring instrument. The mean values of the experimental models were compared to those of the master model using one-sample t-test. The samples of each group at the same site were compared with an independent t-test. For all tests, a significance level of 5% (0.05) was considered. Results: The confidence intervals from M1, M2, and M4 sites for both gypsum and resin models presented statistically lower linear distance when compared to the reference values. At m3, the mean value for the gypsum models was not statistically different from the reference mean value (p > 0.05); however, resin-printed models presented a statistically different mean value (p < 0.05), as well as lower values of linear distance. Conclusions: When compared to gypsum models, resin- printed models differed greatly from the master model, indicating the need for standardizing the printing protocol, for its variables may influence printed models accuracy.

Accuracy of Implant Level Intraoral Scanning and Photogrammetry Impression Techniques in a Complete Arch with Angled and Parallel Implants: An In Vitro Study

(1) Background: Stereophotogrammetry has recently been investigated showing high accuracy in complete implant supported cases but has scarcely been investigated in cases of tilted implants. The aim of this in vitro study was to compare the accuracy of digital impression techniques (intraoral scanning and photogrammetry) at the level of intraoral scan bodies in terms of angular deviations and 3D discrepancies. (2) Methods: A stone master cast representing an edentulous maxilla using four implant analogs was fabricated. The two anterior implants were parallel to each other, and the two posterior implants were at an angulation of 17 degrees. Digital intraoral scanning (DIOS) impressions were taken after connecting implant level scan bodies to the master cast and STL files were exported (n = 15). Digital photogrammetry (DPG) impressions were captured using a PiC Camera after tightening implant level PiC optical markers and STL files were exported (n = 15). Superimposition was carried out by a software for determining the accuracy of both. (3) Results: Significant angular discrepancies (ΔA) and 3D deviations of scan bodies were found among the groups in trueness with lower deviations for the DPG (p value < 0.001). However, trueness within ISBs varied between angular and 3D deviations and outcomes were not specific to determine the effect of implant angulation. In precision, no significant differences were detected within ISBs and among both groups in terms of angular deviation. However, DPG had less deviations than DIOS group in terms of 3D deviations (p value < 0.001). (4) Conclusion: Digital photogrammetry technique conveyed the utmost accuracy in both trueness and precision for the intraoral scan bodies among both impression methods assessed. In addition, implant angulation did not influence the precision of the impression techniques but affected their trueness without explicit conclusions.