Comparison of four rigid registration strategies for three- dimensional facial scans in growing patients with skeletal Class III malocclusion undergoing facemask therapy

Background Three-dimensional (3D) facial scanning enables radiation-free quantification of soft-tissue changes during orthodontic/orthopedic treatment. However, divergent rigid registration strategies may yield discrepant results, particularly in growing children with specific treatment. This study compared the accuracy of four rigid registrations for assessing soft-tissue changes after facemask (FM) therapy in growing skeletal Class III patients. Methods Twenty-eight late mixed/early permanent dentition patients (mean age 8.2 ± 1.3 years) underwent FM therapy. Pre-(T0) and post-(T1) treatment 3D facial scans and CBCTs were acquired. Following CBCT reference registration, four facial scan surface superimposition methods were tested: (1) FF-ICP (full-face iterative closest point); (2) NR (nasal-root region bounded by infraorbital foramina); (3) NFR (NR extended by an equally wide mid-forehead zone); (4) P-ICP (landmark-guided ICP). Outcomes were whole-face/regional root-mean-square error (RMS), subjective visual agreement (number rating scale, NRS 1–10), and computer-vision metrics (SSIM, PSNR, KLD, SID, CS). Results NR achieved the lowest global RMS (0.76 mm), outperforming FF-ICP (1.03 mm), NFR (1.14 mm), and P-ICP (1.24 mm). Except for CS, all computer-vision metrics and the subjective NRS showed that NR was closest to the CBCT reference. Regionally, RMS of lips and chin were higher than other regions across methods; forehead RMS was lowest in all parameters and showed no between-method differences. Conclusions For growing skeletal Class III patients undergoing FM therapy, rigid registration with the nasal-root region (NR)alone yields the more consistent results with CBCT anterior cranial base superimposition, although its combination with forehead region is recommended in growing patients. To achieve precise therapeutic evaluation, rigid 3D facial scan registration requires personalization based on the patient's current growth stage and therapeutic approach.