Reliability and accuracy of smartphone-based facial scanning: Validation with direct anthropometry

To evaluate the reliability and measurement agreement of two smartphone-based facial scanning applications-Qlone (iOS) and Magiscan (Android)-with direct anthropometry following scale factor calibration, and to assess intra- and inter-examiner reliability.

A total of 22 orthodontic patients were enrolled in this prospective study. Nine soft tissue landmarks (midsagittal and bilateral) were marked using 2.5 mm stickers and three vertical and six transverse linear distances were measured using digital vernier calliper (resolution 0.01 mm, accuracy ±0.02 mm) for direct anthropometry and MeshLab software following .obj export and scaling for both applications. Reliability was assessed in 10 randomly selected patients after 15 days.

Repeated-measures ANOVA (F-statistic, 2 and 42 degrees of freedom) showed no statistically significant differences between Qlone, Magiscan, and direct anthropometry for 8/9 parameters (p range 0.165-0.988; η²p range 0.001-0.082). A statistically significant difference was found for Exo Left-N (p = 0.041; η²p = 0.141; mean difference 0.55 mm), not considered clinically significant. Bland-Altman analysis confirmed clinically acceptable agreement (mean bias ⩽ 0.55 mm across all comparisons). Intraclass correlation coefficients (two-way mixed model, absolute agreement) were excellent for all parameters (intra-examiner 0.998-1.000, inter-examiner 0.994-0.999), with relative TEM ⩽ 0.79% and reliability coefficient R ⩾ 0.994.

Following scale factor calibration, smartphone-based 3D facial scanning application demonstrate high reliability, precision, and clinically acceptable agreement with direct anthropometry. These applications offer a portable and cost-effective option for facial assessment in orthodontic practice, pending independent accuracy validation using a non-circular reference standard.