This retrospective observational reader study will evaluate artificial intelligence (AI)-assisted implant planning using anonymized cone-beam computed tomography (CBCT) datasets from patients with complete edentulism or a clinically equivalent edentulous condition. AI-generated implant plans will be compared with expert reference plans created by clinicians using the same CBCT data. The study will assess the clinical acceptability of AI-generated implant plans, geometric agreement with expert plans, anatomical safety, workflow time, and agreement between expert reviewers where applicable. The study uses previously acquired anonymized imaging data and does not involve patient recruitment, treatment allocation, additional imaging, clinical intervention, or prospective follow-up.
This study is designed as a retrospective non-randomized comparative reader study. Anonymized CBCT datasets acquired during routine clinical care will be used for implant planning assessment. For each eligible case, expert clinicians will create reference implant plans without access to AI-generated plans. The AI system will generate implant planning outputs from the same CBCT datasets, and expert clinicians will review the AI-generated plans using a standardized assessment approach. The main evaluation will compare AI-generated plans with expert reference plans within the same case. Outcomes will include clinical acceptability of the AI-generated plan, geometric agreement between AI-generated and expert plans, anatomical safety relative to relevant risk structures, time required for expert planning versus AI-plan review and correction, and inter-reader agreement where applicable. The study does not test an autonomous AI decision-making system. The AI workflow is evaluated as a clinical decision-support tool, and all AI-generated plans are subject to expert clinician review. No new imaging examinations, treatment allocation, patient intervention, or prospective clinical outcome assessment will be performed.
Study Type
OBSERVATIONAL
Enrollment
100
AI-assisted implant planning workflow applied to anonymized CBCT datasets. The workflow generates implant planning outputs for expert review and comparison with expert reference plans. It is evaluated as a clinical decision-support workflow and does not involve patient treatment, additional imaging, or autonomous clinical decision-making.
Pavlov First Saint Petersburg State Medical University
Saint Petersburg, Sankt-Peterburg, Russia
Clinical acceptability of AI-generated implant plans
Proportion of AI-generated implant plans rated by expert clinicians as accepted without modification, accepted after minor modification, accepted after major modification, or rejected.
Time frame: Baseline
Geometric agreement between AI-generated and expert reference implant plans
Geometric agreement will be assessed for matched implants using entry-point deviation, apical deviation, and angular deviation between AI-generated and expert reference implant positions.
Time frame: Baseline
Anatomical safety of AI-generated implant plans
Anatomical safety will be assessed using minimum distances from planned implants to relevant anatomical risk structures and the presence or absence of predefined safe-margin violations.
Time frame: Baseline
Workflow time for AI-assisted planning review compared with expert planning
Time required for independent expert implant planning will be compared with the time required for expert review and correction of AI-generated implant plans.
Time frame: Baseline
Inter-reader agreement for clinical acceptability ratings
Agreement between expert clinicians will be assessed for clinical acceptability ratings of AI-generated implant plans where more than one expert evaluates the same cases.
Time frame: Baseline
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.