AI-based three-dimensional assessment of alveolar bone grafting using iliac crest and mandibular symphysis bone combined with L-PRF in unilateral cleft lip and palate

Abstract

Objectives: This thesis aimed (1) to develop and validate a standardized AI-based 3D workflow for the assessment of alveolar bone grafting (ABG) outcomes and mandibular symphysis (MS) donor-site bone regeneration, and (2) to apply the validated methodology to evaluate the effect of leukocyte-platelet-rich fibrin on alveolar grafting performed with iliac crest (IC) and MS bone. Materials and Methods: Two sequential studies were conducted. First, an AI-powered 3D imaging workflow using automated orientation, segmentation, voxel-based superimposition, and shape correspondence analysis was developed and validated in a sample of 21 patients with cleft lip and alveolus. This workflow was used to assess ABG performed with MS bone, evaluating bone formation at the cleft site and MS donor site bone regeneration. Subsequently, the validated workflow was refined to include volumetric bone quantification and applied in a prospective cohort study of 92 patients with unilateral complete cleft lip and palate who underwent ABG. Patients were allocated in groups according to graft donor site (IC or MS) and the use of L-PRF (with or without). CBCT scans were obtained preoperatively and at a 12-month follow-up. Volumes of maxillary bone formation and MS donor site bone regeneration were assessed using the CBCT scans and 3D models. Shape correspondence analysis was applied to compare pre-and post-harvesting symphyseal morphology. Results: The AI-powered 3D workflow demonstrated feasibility, reproducibility, and clinical applicability for the assessment of ABG outcomes and MS donor site bone regeneration. Volumetric bone formation does not differ significantly regardless to graft source or L-PRF use. No significant main effects of graft type (IC vs MS) or L-PRF use were observed, and no significant interaction between these factors was detected (P > 0.05). MS donor site bone regeneration was high, with a mean regeneration of 92.5 ± 4.8%. Shape correspondence analysis showed a mean residual discrepancy of 2.81 ± 1.99 mm relative to pre-harvesting morphology. All patients were discharged within 24 hours in stable condition, reporting minimal postoperative discomfort and no postoperative complications. Conclusion: The proposed AI-powered 3D imaging workflow is reliable and clinically applicable for assessing ABG outcomes and mandibular symphysis donor site bone regeneration. In this cohort, L-PRF does not significantly influence volumetric bone formation, regardless of graft source, and no interaction between graft type and L-PRF was observed. Bone regeneration at the mandibular symphysis donor site was high, with no permanent complications.