Introduction

Neurological development is a complex and dynamic process in which brain structure, function, and connectivity evolves throughout the first three decades of life. Despite recent quality and resolution advancements in MRI, there is no high-quality paediatric data.

Methods

This pilot study employs short (~8 min) structural acquisition protocols at 0.65mm isotropic. Four T1-w scans were collected for two female participants, aged 9 and 12. Multiple shorter scans facilitate breaks between scanning and thus navigate the issue of participant compliance. Then, using template generation protocols, single participant images are averaged where T1-w and T2-w ‘templates’ are generated for each participant. Our methods are publicly available at: https://osf.io/ckh5t/

Results

Applying template generation methods to single participant datasets reduces head-motion artefacts, reveals structural detail and facilitates up-sampling to a voxel size of 0.5mm isotropic in the final data. Figure 1 compares outputs from our method to the Haskins child atlas, illustrating the marked improvement in structure visibility and an 8-fold increase in resolution we achieve.

Discussion

To our best knowledge this represents the highest resolution pediatric MRI dataset by a significant margin. It permits the comprehensive investigation of all ~1000 brain structures, including challenging structures such as the hippocampal subfields which can only be sufficiently visualised at high resolution. Our methods, tailored to the child brain, are key to advancing paediatric clinical interventions and facilitating the investigation of child brain anatomy directly, instead of deduction from adult anatomy.