A Multi-channel 4D Probabilistic Atlas of the Developing Brain: Application to Fetuses and Neonates

Abstract

Brain atlases are widely used in the neuroscience community as a tool for providing a standard space for comparison of subjects. Most efforts so far have focused on single modality brain atlases, however, the need remains for a multi-channel spatio-temporal atlas to model the dynamic changes in multiple modalities and to provide tissue priors used to enhance the intensity-based classification of brain MRI. In this paper, we present an approach for constructing a 4D multi-channel atlas of the developing preterm brain which incorporates multiple modalities and tissue segmentations. Moreover, the approach is applied to a new dataset of magnetic resonance (MR) images from 80 fetuses aged between 23 and 37 weeks gestational age (GA) at time of scan. The resulting atlases showed to improve atlas-based automatic segmentation comparing to probabilistic atlases generated using affine registration. Also, the fetal atlas provides a natural benchmark for assessing preterm born neonates and give some insight into differences between the groups. The atlases are publicly available at www.brain-development.org.