An Implementation of FEM Simulation for Brain Shift Prediction to Enhance the Neurosurgical Planning

Abstract

In many hospitals, especially in resource-limited settings, neurosurgeons lack access to advanced intraoperative imaging, such as intraoperative MRI or CT. This limitation severely restricts their ability to adapt surgical plans in real-time, especially when anatomical changes occur during the procedure. One of the most critical of these changes is brain shift, a phenomenon where brain tissue deforms during surgery, reducing the accuracy of image-guided interventions. This study addresses that critical gap by introducing a finite element method (FEM)-based simulation that predicts brain shift using only standard preoperative imaging, MRI and CT. Unlike existing approaches that rely on intraoperative imaging or data-driven updates, the proposed method models deformation based on physical factors such as gravity, remaining cerebrospinal fluid, and brain–skull contact, using a static linear elastic formulation. The method is feasible, hardware-efficient, and deployable in typical clinical environments. Validation through phantom experiments confirmed the feasibility and clinical acceptability of the approach, achieving a mean lesion localization error of 1.94±0.59 mm and shift-correction improvement of 46.12±25.84% compared with conventional planning. With a total simulation time of under six minutes, the proposed approach provides a practical and scalable solution to support neurosurgical planning in settings without intraoperative imaging.