A pipeline for interactive cortex segmentation

Abstract

In various clinical or research scenarios, such as neurosurgical intervention planning, diagnostics, or clinical studies concerning neurological diseases, cortex segmentation can be of great value. As, e.g., the visualization of the cortical surface along with target and risk structures enables conservative access planning and gives context information about the patient-specific anatomy. We present an interactive cortex segmentation pipeline (CSP) for T1-weighted MR images, utilizing watershed and level set methods. It is designed to allow the user to adjust the intermediate results at any stage of the segmentation process. Particular attention is paid to the appropriate visualization of the segmentation in the context of the original data for verification and to different interaction methods (manual editing, parameter tuning, morphological operations). Evaluation of the interactive CSP is performed with the Segmentation Validation Engine (SVE) by Shattuck et al. (NeuroImage 45(2):431---439, 2009). The segmentation quality of our method is comparable to the best results of three different established methods: the brain extraction tool (BET), brain surface extractor (BSE), and hybrid watershed algorithm (HWA). Being designed for interaction, the CSP integrates the users' expertise by allowing him to perform correction at any stage of the pipeline, enabling him to easily achieve a segmentation fulfilling his specific needs.