Tracking structural brain changes at 0.05T using densely sampled neuroimaging, autonomous control, and super-resolution reconstruction - SUMMARY This study aims to demonstrate accurate quantification of changes in brain morphology at 0.05T. We plan to accomplish this tracking by exploiting the portability of our 0.05T scanner to acquire dense temporal sampling of neuroimaging data coupled with autonomous control and deep-learning-based super-resolution reconstruction. Magnetic Resonance Imaging (MRI) has been the primary tool to investigate these disorders, highlighting structural, functional, and metabolic changes. For example, 3T structural MRI data have significantly improved our understanding of the youths’ developing brain in health and disease. Recent studies have highlighted the need for densely sampled temporal neuroimaging data to maximize clinical insight in patients with mental health challenges. High-field systems' cost, power, and siting requirements impede dense longitudinal imaging in large populations, especially in low-resource settings. In contrast, we hypothesize that a portable 0.05T can deliver the required densely sampled neuroimaging data. We expect equivalent statistical power in detecting brain changes associated with young adults at 0.05T with a five-time-point than a two-time-point 3T study. However, 0.05T MRI suffers from lower spatial resolution and signal-to-noise ratio (SNR), impacting the volumetric accuracy required to monitor brain changes using structural imaging. These limitations render these scanners supplementary devices to high-field systems. For meaningful use, there is a critical need to develop novel methods to produce low-field, structural MRI data statistically equivalent to or better than 3T data. To address this gap, we hypothesize that three orthogonal acquisitions and using DL-based super-resolution reconstruction of five-timepoint data from children (10 – 17 years) at 0.05T provide similar accuracy to two-timepoint 3T data. We have demonstrated the image quality of 0.05T data in phantoms over thirty sessions and the pipeline for in vivo data acquisition, reconstruction, and segmentation on one healthy volunteer. These data show the potential of 0.05T data for brain volumetry and the benefits of consistent scanner operation and image quality. In Aim 1, we will demonstrate the feasibility of our 0.05T scanning pipeline to track brain changes and address the need for the high-density temporal sampling of neuroimaging data. Second, the deployment of 0.05T is affected by artifacts caused by external magnetic interference, magnetic field inhomogeneity, subject motion, degrading image quality, and scanner operation. To address these challenges, we will adopt our automated MRI methods to 0.05T in Aim 2. The densely sampled temporal neuroimaging data will yield early, accurate, and personalized developmental trajectories of the youth brain, with the potential to scale this project in an equitable and accessible manner. These methods will enable the broader scientific community to perform previously prohibitive structural MRI studies without compromising accuracy, especially in underserved settings.
