Evaluating analytic strategies to obtain high-resolution, vertex-level measures of cortical neuroanatomy in children in low- and middle-income countries.

High-field magnetic resonance imaging to explore brain structure and function remains limited to high-resource settings. Novel, low-field (<0.1 T) imaging offers a more cost-effective/accessible alternative. However, the validity of low-field data at spatial resolutions relevant to research and clinic (vertex-level) remains unclear. Hence, we examine paired high-field (reference) and low-field (single/multi-orientation scans processed through established/novel pipelines) data (12 children [10-12 yrs] in a low- and middle-income country [LMIC]). We assess high-field/low-field correspondence between vertex-level measures of cortical volume, surface area, and cortical thickness; and compare analytic strategies. High/low-field images show weak-to-moderate global correspondence (cortical volume, surface area: Pearson's r ≤ 0.6, cortical thickness r ≤ 0.3), and weak-to-very strong local correspondence (r ≤ 0.99). Greatest correspondence is achieved with multi-orientation images and a pipeline adjusted for low-resolution images (recon-all-clinical); or image enhancement (SynthSR) plus standard processing (FastSurfer); but agreement varies across brain based on input, analytic strategy, and neuroanatomical feature. We provide an application to interactively explore our results. Thus, low-field imaging can provide reliable, high-resolution estimates of cortical volume and surface area, but not cortical thickness; and analytic approaches should be selected based on multiple considerations. Once validated, this research may help deploy low-field imaging to aid research/evidence-based clinical work in high- and low-resource settings, including LMIC.