To compare the scans of different intraoral scanners (IOSs) and laboratory scanners (LBSs) to those of an industrial-grade optical scanner by measuring deviations of complete-arch implant-supported frameworks from their virtual design file. Ten polyetheretherketone (PEEK) and 10 titanium (Ti) complete-arch implant-supported frameworks were milled from a master standard tessellation language (STL) file. An industrial-grade blue light scanner (AT), 2 LBSs (MT and E4), and 3 IOSs (PS, T3, and T4) were used to generate STL files of these frameworks. All STLs were imported into an analysis software (Geomagic Control X) and overall root mean square (RMS) values were calculated. Marginal surfaces of all STL files were then virtually isolated (Medit Link v 2.4.4) and marginal RMS values were calculated. Deviations in scans of tested scanners were compared with those in scans of AT by using a linear mixed effects model (α = 0.05). When the scans of PEEK frameworks were considered, PS and T3 had similar overall RMS to those of AT (p ≥ .076). However, E4 and T4 had higher and MT had lower overall RMS than AT (p ≤ .002) with a maximum estimated mean difference of 13.41 µm. When the scans of Ti frameworks were considered, AT had significantly lower overall RMS than tested scanners (p ≤ .010) with a maximum estimated mean difference of 31.35 µm. Scans of tested scanners led to significantly higher marginal RMS than scans of AT (p ≤ .006) with a maximum estimated mean difference of 53.90 µm for PEEK and 40.50 µm for Ti frameworks. Only the PEEK framework scans of PS and T3 led to similar overall deviations to those of AT. However, scans of all tested scanners resulted in higher marginal deviations than those of AT scans. Scans performed by using PS and T3 may be alternatives to those of tested reference industrial scanner AT, for the overall fabrication trueness analysis of complete-arch implant-supported PEEK frameworks.

The aesthetic component (AC) of the Index of Orthodontic Treatment Need (IOTN) is internationally recognized as a reliable and valid method for assessing aesthetic treatment need. The objective of this study is to use artificial intelligence (AI) to automate the AC assessment. A total of 1009 pre-treatment frontal intraoral photos with overjet values were collected. Each photo was graded by an experienced calibration clinician. The AI was trained using the intraoral images, overjet, and two other approaches. For Scheme 1, the training data were AC 1–10. For Scheme 2, the training data were either the two groups AC 1–5 and AC 6–10 or the three groups AC 1–4, AC 5–7, and AC 8–10. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were measured for all approaches. The performance was tested without overjet values as input. The intra-rater reliability for the grader, using kappa, was 0.84 (95% CI 0.76–0.93). Scheme 1 had 77% sensitivity, 88% specificity, 82% accuracy, 89% PPV, and 75% NPV in predicting the binary groups. All other schemes offered poor tradeoffs. Findings after omitting overjet and dataset supplementation results were mixed, depending upon perspective. We have developed deep learning-based algorithms that can predict treatment need based on IOTN-AC reference standards; this provides an adjunct to clinical assessment of dental aesthetics.

To evaluate the suitability of intraoral scanners (IOSs) to analyze the fabrication trueness of titanium complete-arch implant-supported frameworks by comparing with an industrial-grade scanner and investigate how anti-reflective scan powder affects measured deviations. Ten titanium complete-arch implant-supported frameworks were milled from a reference standard tessellation language (STL) file. An industrial-grade blue light scanner (ATOS Core 80 (AT)) and three IOSs (Primescan (PS), TRIOS T3 (T3), and TRIOS T4 (T4)) with (PS-P, T3-P, and T4-P) or without (PS, T3, and T4) anti-reflective scan powder application were used to generate test STL (TSTL) files of the frameworks. Reference STL and TSTLs were imported into a metrology-grade analysis software (Geomagic Control X) and whole surface root mean square (RMS) values were calculated. Another software (Medit Link v 2.4.4) was used to virtually isolate marginal surfaces of all STL files and marginal RMS values were calculated by using the same metrology-grade analysis software. A linear mixed effects model was used to compare the transformed deviations of the scans performed by using each IOS (with or without powder) with the deviations of those performed by using the reference AT scanner within each surface, where a Box-Cox type transformation was used for variance stability. Bonferroni corrected post-hoc tests were used to compare conditions within each IOS (α=0.05). All IOSs had significantly higher whole surface and marginal RMS values than AT, regardless of the condition (P≤.002). However, scan powder application did not affect the whole surface and marginal RMS values in scans of tested IOSs (P≥.054). Measured whole surface and marginal deviations in all IOS scans performed with or without the use of scan powder were higher than those in AT scans. The application of anti-reflective scan powder did not affect the deviations in scans of tested IOSs. Even though deviations measured in the scans of tested scanners were significantly different than those in the reference scanner, the maximum raw mean difference was 37.33 µm and the maximum raw confidence interval value of estimated differences was 47.88 µm, which can be considered clinically small taking into account the size of the frameworks tested. Therefore, tested intraoral scanners may be feasible to scan prostheses similar to or smaller than tested frameworks for fabrication trueness analysis, which may facilitate potential clinical adjustments.

The expectation–maximization (EM) algorithm, often used for maximum likelihood estimation, has not seen much application in addressing high-dimensional regularization challenges within linear mixed-effects models. This study introduces the EMLMLasso algorithm, which merges the EM algorithm with the widely used and efficient R package glmnet, enabling Lasso variable selection for fixed effects in such models. We thoroughly evaluate its performance, comparing it to two existing algorithms from R packages glmmLasso and splmm. Our findings, based on simulations and real-world cases, demonstrate the robustness and effectiveness of our approach, even when the number of predictors (p) exceeds the number of observations (n). Notably, across most scenarios, the EMLMLasso algorithm consistently outperforms both glmmLasso and splmm. Moreover, our method is versatile and straightforward to implement, with the potential for extensions to include ridge and elastic net penalties in linear mixed-effects models.