The CRL is engaged in the development of techniques for generating quantitative biomarkers of disease from newborn brain magnetic resonance imaging (MRI). This is an end-to-end effort beginning with the acquisition of high quality MR images and ending with novel post- acquisition processing methods for extracting statistically useful information from MR scans. Ongoing research focuses on the following areas:
Optimization of clinical and research pulse sequence parameters for newborn imaging at 3 Tesla
We are currently acquiring high-resolution T1-weighted and T2- weighted structural images along with high-angular direction (35) diffusion tensor imaging. These sequences have been optimized for the unique size and composition of the smallest newborn brains. Work is ongoing to optimize pulsed arterial spin labeling (PASL) MRI in order to generate a quantitative map of perfusion which can be used to study brain injury related to failures of perfusion.
Additionally, we are investigating techniques for motion-robust MRI to aid in imaging of uncooperative, unsedated neonates. We routinely perform "feed and wrap" imaging of sleeping, unsedated newborns with a high success rate.
Automatic measurement of MRI biomarkers
We have developed a fully-automated, comprehensive analysis pipeline for generating quantitative tissue volumes from newborn MRI. This work includes efforts in post-acquisition artifact correction and noise filtering, inter- and intra-subject registration, the construction of gestational-age specific atlases, the removal of extra-cerebral tissue from MR scans (aka "skull stripping"), optimal statistical classification and related segmentation or image-labeling techniques.