Facial Landmark Guided Accelerated Block Matching Two-Stage Point Cloud Registration Method for Robot-Assisted Stereotactic Navigation System

Le résumé fourni par la source

Depth camera-based point cloud acquisition is being increasingly integrated into frameless stereotactic navigation systems to support broader clinical deployment. However, intraoperative point clouds obtained in this way often suffer from low spatial resolution, significant morphological discrepancies, and limited overlap, collectively complicating registration and impeding accurate localization of intracranial lesions. To address these limitations, this study proposes a facial landmark guided accelerated block matching two-stage registration method (FLABM-2R), where facial landmark extraction and alignment mitigate morphological differences and enable rapid initial alignment of the patient head point clouds, followed by accelerated semi-global block matching for fast refined registration of low-overlap head point clouds. Integrated into a robot-assisted stereotactic navigation system, the proposed method was extensively evaluated through software and hardware experiments, achieving high accuracy and robustness. By removing the requirement for fiducial markers and manual surgeon intervention, this approach enhances procedural efficiency while lowering infection risk in stereotactic neurosurgery.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.