Figure 2. . Schematic of network construction in native and template space for a single subject. Native space HCP HARDI data was reconstructed using GQI and CSD for each subject. Deterministic tractography was performed separately for the GQI and CSD reconstructed native space data. The native space data was reconstructed using CSD to fairly compare the impact of FODR. The scale 60 Lausanne parcellation was applied to assign nodes to the streamlines. The structural connectivity matrix was weighted using streamline count. For FODR, the native space FODs reconstructed using CSD were reoriented according to the deformation field output by ANTs. The subject’s native space parcellation was warped into the template space by using the same deformation field output by ANTs. The raw HARDI data was reoriented as well as the b-vectors to reconstruct ODFs in template space for QSDR. The subject’s native space parcellation was warped into the template space by using QSDR’s internal mapping. Deterministic tractography was performed for FODR and QSDR after normalization. The native space streamlines generated from ODFs reconstructed using GQI were directly warped into template space by using the deformation field output by ANTs. Node assignment and network construction for FODR, QSDR, and DSN follow the same workflow as in native space. After the connectivity matrices are constructed, the impact of different spatial normalization approaches can be measured by comparing the similarity of the connectivity matrices and network metrics derived from them for the same subject before and after normalization.