Fig. 5.

Comparison of the differences in spatial normalisation between the FEAT and dHCP pipelines in a representative subject. Top row: the week 40 GA standard template in grey-scale, with a tissue boundary overlaid in green to aid assessment of registrations. The boundary is taken from the same atlas as the standard template, and includes the GM-WM boundary in the cerebrum, as well as the outer boundary of the brainstem and cerebellum. Second row: the mean functional image registered to standard space using the FEAT pipeline registrations (FSL's FLIRT and FNIRT). Third row: the same mean functional image registered to standard space using the dHCP pipeline registrations (FSL's FLIRT-BBR and ANTs's SyN). In general, the FEAT registrations tend to incorrectly register the GM-CSF boundary of the functional image to the GM-WM boundary of the template, likely due to lack of BBR, and this is corrected in the dHCP result. Also, the cerebellum and brainstem are more accurately aligned with the template in the dHCP result. In this specific subject, several other improvements are visible in the dHCP results, with example regions in each view highlighted with a red circle.