A Novel AI-Assisted 3D Hybrid Soft and Hard Tissue Model for Peri-implant Diagnosis and Treatment Planning: A Proof-of-Concept Study of Immediate Implant Planning in the Anterior Maxilla.

OBJECTIVES

Immediate implant placement in the maxillary esthetic zone requires precise three-dimensional (3D) positioning within compromised sockets. Conventional planning platforms integrate cone-beam computed tomography (CBCT) and intraoral scanning (IOS) data but exclude critical parameters such as gingival phenotype and thickness. This study aimed to develop and evaluate a novel patient-specific virtual 3D peri-implant model (PIM) that converts AI-segmented dental hard- and soft-tissue data into a realistic 3D soft- and hard-tissue model for peri-implant diagnosis and treatment planning, fusing AI-segmented CBCT, IOS, and gingival tissue data into a single implant planning framework.

METHODS

Thirty patients requiring tooth extraction and immediate implant placement of a maxillary anterior tooth were retrospectively included. CBCT and IOS datasets were processed through an AI-driven platform to segment teeth, alveolar bone, gingiva, and surrounding structures, which were then integrated into a patient-specific 3D virtual model. Virtual tooth removal created a simulated post-extraction socket, allowing prosthetically driven implant planning. Three specialist clinicians from oral surgery, prosthodontics, and periodontics independently performed implant planning under two conditions: conventional CBCT and IOS alone, and PIM-assisted planning. Inter-operator variability and agreement were evaluated across coronal, apical, and angular dimensions using continuous and ordinal outcome measures.

RESULTS

PIM workflow feasibility (successful model generation and implant planning) was 100% (30/30 cases). PIM was associated with lower mean inter-operator deviations for apical (Holm-adjusted p = .009) and angular positioning (Holm-adjusted p < .001), but not coronal positioning (Holm-adjusted p = .10). Major angular deviations decreased from 40.0% to 6.7% of implants (Holm-adjusted p = .028), with no categorical difference for coronal or apical deviation.

CONCLUSION

Converting AI-segmented dental hard- and soft-tissue data into a realistic hybrid 3D tissue model was feasible and significantly reduced mean apical and angular inter-operator deviations, with the ordinal benefit most evident for angular positioning. These findings support further prospective evaluation of the PIM workflow.

CLINICAL SIGNIFICANCE: By fusing bony and gingival characteristics into a single realistic 3D digital tissue model, PIM may support more informed decisions regarding implant position, grafting, abutment selection, emergence profile design, and restorative planning before surgery.