44 Chapter 2 Ground truth measurement of CBF was not present in this study to ultimately prove the validity of the proposed method for spatial normalization. However, since all methods contained the same preprocessing except for the spatial normalization, we assume that a difference in extracted regional CBF is a sufficient measure of the difference between the methods. Moreover, we assume that the method that performed better at the qualitative and quantitative evaluation of the spatial normalization will also be closer to the ideal normalization. As the registration of spatial MRI sequences to structural regT1 images of the brain is a major challenge in patients with unmyelinated brains (such as neonates), there are other example studies concerning the use of contrast that is not usual for this population. Because many immature white matter tracts cannot be differentiated with conventional MR images (T1w or T2w) caused by insufficient myelination during the preterm development period, Feng et al. and Yoshida et al. used DTI as a substitute for T1w in a preterm population.4, 5 Another example, which is also relevant for neonates, is tract based spatial statistics, permitting a voxel-wise statistical analysis of the entire white matter skeleton instead of the usual DTI ROI approach.24, 25 This study could be an initial impetus for future large cohort pediatric ASL studies using ASL as an unusual contrast. Imaging of preterm and term-born infants for clinical indications is of great interest. It has been demonstrated that various injuries, due to perinatal risk often leads to damage in selective white matter.26 De Vis et al. have already demonstrated that CBF progressively increases during the first period of life as synaptogenesis, myelination, and brain functional activity progress.27 Miranda et al. have observed cerebral perfusion was significantly higher in preterm newborns studied at term-corrected age than in term-born newborns, indicating that brain perfusion may be influenced by developmental and postnatal age.28 Both studies were based on manual ROI delineations of cortical regions or on manual placement of control points in GM and WM. Using the proposed normalization method provides clinical users an automatic delineation, which could potentially lead to a higher reproducibility and regional accuracy. Limitations One main limitation is that we tested whole-brain alignment only, i.e., we have not systematically tested the alignment quality on the level of individual cortical structures. While the method proved to provide usable or excellent spatial normalization and thus should allow the reliable assessment of perfusion in specific regions of the MNI space, a more thorough validation is needed to demonstrate whether this normalization method allows regional CBF evaluation in MNI space using gray and white matter masks. Currently, the method thus allows the study of regional brain perfusion even in the absence of a T1w image, but further validation is needed to fully replace T1w
RkJQdWJsaXNoZXIy MTk4NDMw