An acquisition and reconstruction framework to enable mesoscale human fMRI on clinical 3 Tesla scanners

项目总结/文摘

PROJECT SUMMARY/ABSTRACT Functional MRI (fMRI) is the most widely-used tool to noninvasively measure brain function and has produced much of our current knowledge about the functional organization of the human brain. However, all fMRI methods measure neuronal activity indirectly by tracking the associated local changes in blood flow, volume and oxygenation, which limit their spatiotemporal specificity to the underlying neuronal activity. While this is often viewed as the fundamental limitation of fMRI, recent optical imaging studies in animal models have shown a tight coupling between microvascular diameter changes and neural activity. These data indicate that human fMRI— as it is performed today—has vast untapped potential that can only be reaped if our measurements can be made sensitive exclusively to changes in these smallest blood vessels. Recent human studies have demonstrated that fMRI based on tracking changes in cerebral blood volume (CBV) with high sensitivity to microvascular diameter changes indeed provide improved neural specificity compared to the conventional blood-oxygenation-level- dependent (BOLD) method. Since the advent of fMRI, BOLD has been the most used fMRI contrast due to its robustness and high sensitivity, yet consensus is building that CBV provides far more faithful measurements of neural activity. Adoption of this powerful non-BOLD fMRI approach has been lacking due to its low sensitivity. To address this, we will develop new imaging methods to dramatically increase sensitivity of CBV-based fMRI. The key to our approach is the recognition that it is now possible, with advanced acquisition methods that we have recently developed, to separate “contrast encoding” in fMRI from image encoding. Because the standard image encoding with EPI in fMRI inherently introduces T2* weighting (and thus BOLD contrast), emerging non-BOLD techniques must inefficiently acquire two sets of data for every measurement to remove the unwanted BOLD contamination in post-processing. The need to acquire two sets of images, along with the necessary post-processing, plus the T2* signal loss combine to cause up to 4× SNR-efficiency loss. Our method, based on our distortion- and blurring-free Echo-Planar Time-resolved Imaging (EPTI) technology, overcomes this SNR loss by eliminating the unwanted BOLD weighting. We call our new framework “Mz fMRI” as it provides a means to generate fMRI contrast based purely on longitudinal magnetization (Mz), applicable to various non- BOLD fMRI methods. We will provide a proof of concept of this powerful framework by integrating EPTI with “VASO” to create an efficient CBV-fMRI without distortion, blurring, or the need for BOLD removal. Our simulations indicate that our approach will deliver sufficient sensitivity for sub-millimeter CBV-fMRI at 3T, and will perform better than existing CBV methods at 7T; the availability of these powerful methods at 3T will open non-BOLD fMRI up to the entire fMRI community, boosting neuronal specificity and enabling broad application of mesoscale fMRI—such as studies of cortical columns and layers and small subcortical nuclei— and thereby opening new possibilities for mapping whole-brain circuitry.