Neuroimage as a Biomechanical Model: Toward New Computational Biomechanics of the Brain

Abstract

In recent years, predicting brain deformations during surgery using methods of computational biomechanics has become a viable alternative to purely image-based techniques. However, the difficulties with patient-specific computational grid generation prevent the widespread application of biomechanical modeling in medicine. For more efficient computational grid generation, we propose a statistical meshless model based on fuzzy tissue classification and mechanical property assignment, and meshless (i.e., based on the unstructured cloud of points that do not form elements) solution method. Instead of hard segmentation that divides intracranial area into nonoverlapping, constituent regions we use statistical classification to get the fuzzy membership functions of tissue classes for each voxel. Material properties are assigned to integration points based on this soft classification. Verification example shows that the proposed model gives equivalent results—difference in computed brain deformations of at most 0.2 mm—to the finite element method (FEM) and can certainly be considered for use in future simulations. Based on this concept, patient-specific computational models can be more efficiently and robustly generated in the clinical workflow.