Optimization and comparative evaluation of nonlinear deformation algorithms for atlas-based segmentation of DBS target nuclei

• DOI: 10.1016/j.neuroimage.2018.09.061
• Publication Date: 2018-09-26T21:55:13Z
• Authors: Siobhan Ewert; Andreas Horn; Francisca Finkel; Ningfei Li; Andrea A. Kühn; Todd M. Herrington
• Keywords: Subthalamic Nucleus; Human Connectome Project; Spatial normalization

Nonlinear registration of individual brain MRI scans to standard templates is common practice in neuroimaging and multiple algorithms have been developed refined over the last 20 years. However, little has done quantitatively compare available much that work exclusively focused on cortical structures given their importance fMRI literature. In contrast, for clinical applications such as functional neurosurgery deep stimulation (DBS), proper alignment subcortical between template space important. This allows atlas-based segmentations anatomical DBS targets subthalamic nucleus (STN) internal pallidum (GPi). Here, we systematically evaluated performance six modern established normalization segmentation results by calculating 11,000 nonlinear warps 100 subjects. For each algorithm, its using T1-or T2-weighted acquisitions alone or a combination T1-, T2-and PD-weighted parallel. Furthermore, present optimized parameters best performing algorithms. We tested algorithm two datasets, state-of-the-art cohort young subjects age- MR-quality-matched typical Parkinson's Disease cohort. Our final pipeline able segment with precision comparable manual expert both cohorts. Although study focuses prominent targets, STN GPi, these methods may extend other small like thalamic nuclei accumbens.