AI-Driven 3D Reconstruction of Anatomical Variations for Personalized Surgical Plan

Abstract

Background: This study was conducted to assess the AI-Driven 3D Reconstruction of Anatomical Variations for Personalized Surgical Plan.

Material and methods: Ten patients undergoing CT/MRI were enrolled, and imaging data were processed using an AI-driven 3D reconstruction pipeline. Deep learning–based segmentation enabled generation of patient-specific anatomical models, highlighting variations. Reconstructions were validated by radiologists and surgeons for accuracy and integration into surgical planning.

Results: The system accurately identified diverse anatomical variations, including vascular, sinonasal, and skull base anomalies in 9/10 patients. Mean reconstruction time was 18.6 minutes with a Dice similarity coefficient of 92.4%. High interobserver agreement (κ = 0.87) confirmed clinical reliability for personalized surgical planning.

Conclusion: AI-driven 3D reconstruction proved to be accurate and efficient in detecting anatomical variations with strong expert validation. Its integration into surgical planning shows promise for enhancing precision and personalization in patient care.