Validation of Consumer-Grade 3D Scanners for Clinical Facial Reconstruction: Comparison with a Professional Structured-Light Reference System

Three-dimensional facial scanning is an important tool in reconstructive surgery and burn medicine for objective documentation, treatment planning, and fabrication of patient-specific devices. While professional structured-light scanners provide high geometric accuracy, their cost limits routine implementation. Several low-cost consumer-grade alternatives are available; however, their accuracy for clinically relevant facial applications remains insufficiently validated. Thirty volunteers were initially recruited for facial scanning. Two participants withdrew from further scanning for personal reasons, resulting in a main study cohort of 28 participants. Complete datasets from all 28 participants were available for Artec Eva, Revopoint MIRACO, and Creality CR-Scan 01, whereas 22 Kiri Engine datasets met the predefined quality criteria for smartphone photogrammetry analysis. Facial scans acquired using Revopoint MIRACO, Creality CR-Scan 01, and Kiri Engine were compared with a professional structured-light scanner (Artec Eva) used as the reference system. Six clinically relevant inter-landmark distances were analysed together with global surface deviation relative to Artec Eva reference using Cloud-to-Mesh analysis. Agreement was evaluated using mean signed differences, mean absolute error (MAE), intraclass correlation coefficients (ICC), Bland–Altman analysis, and post-hoc statistical testing. Both MIRACO and CR-Scan demonstrated favourable geometric agreement with the Artec Eva reference. Global surface RMSE relative to Artec Eva reference was 0.72 ± 0.24 mm for MIRACO and 0.57 ± 0.17 mm for CR-Scan, compared with 1.41 ± 0.58 mm for Kiri Engine. More than 90% of facial surface points were within 1 mm of the reference model for MIRACO (91 ± 8%) and CR-Scan (93 ± 7%), whereas Kiri Engine achieved 68 ± 16%. Pooled ICC values ranged from 0.998 to 0.999 across the evaluated systems. Pairwise comparisons showed no significant difference between MIRACO and CR-Scan (p = 0.42), while both significantly outperformed Kiri Engine (p < 0.001). Consumer-grade structured-light scanners demonstrated submillimeter global surface deviations, supporting further investigation for facial surface documentation and monitoring applications. Smartphone photogrammetry showed substantially larger deviations and may be less suitable for applications requiring precise quantitative measurements. Low-cost structured-light systems demonstrated promising geometric performance in healthy volunteers; however, further validation in relevant clinical populations is required before their applicability to reconstructive surgery and burn care workflows can be established.