Figure 11:

Amount of variance in clinical outcome explained when applying various preprocessing options (retrospective 1.5T cohort). A) an exemplary “standard neuroimaging” approach with a monospectral (T1 only) FSL FNIRT based registration and a binary VTA. B & C) Default pathways of Lead-DBS, registering preoperative data into template space in multispectral fashion according to the most optimal method as delineated in (Ewert et al. 2018a). In these approaches, the overlap sum between the E-Field gradient magnitude inside the STN was calculated. D) Using results from C as weighted seeds to isolate fibers from the normative PPMI connectome (table 8), correlating weighted numbers of streamlines to SMA to explain clinical motor improvement.

First column of scatterplots shows direct correlation between VTA-STN overlap (weighted E-field overlap or weighted streamlines to SMA) with clinical improvement. Second column shows GLM with additional clinical covariates. Table 9 shows results for additional preprocessing strategies.