Tetrahedral mesh generation for non-rigid registration of brain MRI: Analysis of the requirements and evaluation of solutions

Abstract

The application we target in this paper is the registration of pre-operative Magnetic Resonance Imaging with the scans acquired intra-operatively during image guided neurosurgery. The objective of this application is improved tracking of tumor boundaries and surrounding brain structures during open skull tumor resection. We focus on a validated, physics-based non-rigid registration approach, which has been used in clinical studies for the last three years. This approach requires tetrahedral tessellation of the brain volume for biomechanical model construction. The analysis of the requirements and available methods to construct such a discretization is the objective of our paper. The paper presents a number of practical contributions. First, we survey the proposed approaches to tetrahedral mesh generation from medical image data. Second, we analyze the application-specific requirements to mesh generation. Third, we describe an end-to-end procedure of tetrahedral meshing for this application using off-the-shelf non-commercial software. Finally, we compare the performance of the considered mesh generation tools in the application context using generic and application-specific quantitative measures.