33 Using a new technique of ASL | Trigonocephaly INTRODUCTION Spatial normalization is an important step for brain image processing; it not only enables group analyses but is also required for automatic segmentation of tissue type and brain regions. Functional or physiological MRI acquisitions, such as arterial spin labeling (ASL) perfusion MRI, typically perform nonlinear registration via conventional structural – mostly T1-weighted (T1w) – scans for their higher resolution and structural contrast. However, in situations where the tissue contrast is low and changing, such as in early phases of myelination in newborns and infants, these structural reference scans may not help or even fail normalization.1-3 The use of other images with higher tissue-contrast could help registration. As an alternative to spatial normalization via segmentation and registration of structural images, studies use contrast from different MRI modalities. Feng et al. used Diffusion Tensor Imaging (DTI) as a substitute for T1w scans.4 In DTI images, premyelination is encountered prior to being detectable at T1w or T2w imaging.5 Similarly, Mutsaerts et al. used cerebral blood flow (CBF) and pseudo-CBF, created from a gray matter (GM) map from segmented T1w image to register individual ASL and T1w volumes instead of using the morphological images for the registration, for example, the ASL control images or M0 scans registered to T1w images in elderly subjects.6 This approach was especially important in cases where the image contrast difference between GM and white matter (WM) was low in ASL control images or in M0 scans, due to, for example, use of strong background suppression or short TR, respectively. This approach can be potentially extended to direct spatial normalization of ASL to standard space in the pediatric population as ASL studies of the brain show sufficient CBF contrast between GM and WM already in early age despite the potential lack of GM/WM contrast in T1w images.5 The problem with spatial normalization in subjects with ongoing myelination is of particular clinical relevance in imaging babies with inborn or acquired disorders impairing normal brain development. Craniosynostosis, referred to the premature fusion of the skull sutures leading to skull and brain deformations, is an example of such disease.7-10 Trigonocephaly, a non-syndromic type of craniosynostosis that presents within a sliding scale of severity in phenotype and brain imaging, is one of the key components in the decision making for surgical treatment in the first years of life. Imaging of these newborns is essential, and ASL is an MRI technique that could provide cerebral perfusion measurements on both a global and regional level. Spatial normalization is then necessary to be able to evaluate perfusion in predefined anatomical regions. However, automatic methods for spatial normalization are challenging in young children with craniosynostosis as there are issues with low GM/ WM contrast and skull deformity, as explained above.7, 11 In previous ASL studies in 2
RkJQdWJsaXNoZXIy MTk4NDMw