Human brain development begins in utero third week post conception, progresses rapidly throughout gestation, and protracts beyond birth. The newborn brain consists of billions of neurons, intricately connected to form a massive neural network that is adaptive and responsive to stimuli from the outside world. Interaction with the world drives dynamic brain development to acquire for instance key language and motor skills.
In the first year of life, the brain doubles in size and develops heterogeneously in space according to a function-dependent schedule. For example, the cerebellum, key to movement control and coordination, doubles in size in the first three postnatal months1, allowing the learning of body manoeuvring through motor babbling. In contrast, the hippocampus, involved in memory consolidation, becomes only 47% bigger1 in the same period. This relatively snail-paced growth of the hippocampus is in line with why childhood memories are typically frail. In addition to size, the brain also evolves regionally in shape and content. The continuous remodelling of the baby brain, stimulated by environmental interactions, ultimately shapes cognition and behaviour.
An essential step toward quantifying early brain development is to create a detailed atlas of brain anatomy as a spatial reference for consolidating and comparing brain features from multiple individuals. An impediment to this endeavour is the dearth of high-quality infant MRI data, curtailing the spatiotemporal details that can be captured in an atlas. Moreover, the types of features captured in existing atlases are typically limited, depicting only partial aspects of brain development and hindering cross examining of complementary features. To address these problems, we created month-to-month brain atlases of infants 2 weeks to 2 years of age, integrating features of cortical surfaces and image volumes in a common space. Our joint surface-volume atlases capture salient traits of the early developing brain and lay the foundation for concurrent analysis of different features across multiple scales in a common coordinate framework.
To construct surface-volume atlases at each month for the first two postnatal years, we processed the high-quality longitudinal magnetic resonance imaging (MRI) data of 37 infants recruited as part of the UNC/UMN Baby Connectome Project2 (BCP) to obtain cortical surfaces (white and pial) and tissue segmentation maps delineating gray matter, white matter, and cerebrospinal fluid. We first constructed the reference surface-volume atlas at year 1 by simultaneous spatial normalization3 and averaging of the cortical white and pial surfaces and image volumes (tissue maps and T1-weighted and T2-weighted images). This reference atlas was then propagated to each month between 2 weeks and 2 years for a complete set of longitudinal surface-volume atlases.
Our infant brain atlas (IBA) at each month faithfully captures the sulcal and gyral features of primary, secondary, and tertiary cortical folds (Figure 1), with developmental timing and left-right lateralization coherent with prior knowledge on early brain development. Our atlases preserve spatiotemporally heterogeneous trajectories of detailed folding and tissue features of the developing brain. Primary visual and motor cortices more developed during infancy experience less changes, in contrast to higher-order cortices that undergo delayed functional development associated with for example decision making, critical thinking, and reasoning.
IBA will facilitate the creation of growth models of healthy infants to pinpoint aberrations due to neuropsychiatric disorders. Our atlases can be leveraged for aggregating information from multiple imaging modalities, such as diffusion and functional MRI, to give a more complete picture of brain development. These atlases (https://iba.yaplab.io) will be an indispensable resource for brain researchers interested in understanding the most rapid and arguably the most influential phase of human brain development.
 Holland, D. et al. Structural growth trajectories and rates of change in the first 3 months of infant brain development. JAMA Neurol. 71, 1266–1274 (2014).
 Howell, B.R. et al. The UNC/UMN Baby Connectome Project (BCP): An overview of the study design and protocol development. Neuroimage. 185, 891–905 (2019).
 Ahmad, S. et al. Surface-constrained volumetric registration for the early developing brain. Med. Image Anal. 58, 101540 (2019).