Advances in High-Field BOLD fMRI

Abstract

This review article examines the current state of BOLD fMRI at a high magnetic field strength of 7 Tesla. The following aspects are covered: a short description of the BOLD contrast, spatial and temporal resolution, BOLD sensitivity, localization and spatial specificity, technical challenges as well as an outlook on future developments are given. It is shown that the main technical challenges of performing BOLD fMRI at high magnetic field strengths-namely development of array coils, imaging sequences and parallel imaging reconstruction-have been solved successfully. The combination of these developments has lead to the availability of high-resolution BOLD fMRI protocols that are able to cover the whole brain with a repetition time (TR) shorter than 3 s. The structural information available from these high-resolution fMRI images itself is already very detailed, which helps to co-localize structure and function. Potential future applications include whole-brain connectivity analysis on a laminar resolution and single subject examinations.

Keywords: 2D EPI; 3D EPI; 7 Tesla; BOLD fMRI; BOLD response; high field.

Figure 1

Schematic of the BOLD hemodynamic response to a brief stimulus at time zero. After the elusive “initial dip” that may arise as a result of initial oxygen uptake before hemodynamic changes occur, the blood flow effect dominates and causes the positive main BOLD response to peak after approximately 4–5 s. The return to baseline is typically preceded by a post-stimulus undershoot which can be of considerable duration.

Figure 2

Hemodynamic effects contributing to the BOLD signal during activation. Panel (A) shows the situation in the baseline state, with only some of the delivered oxygen being utilized. The diagrams at the bottom illustrate the effect of oxygen metabolism (C), blood flow (D) and blood volume (E) increases on deoxyhemoglobin concentration. Blood flow increase dominates, leading to a net reduction of deoxyhemoglobin (B) and an increase in MR signal is hence observed.

Figure 3

Suitable echo-planar imaging (EPI) sampling schemes for high-field fMRI. Panel (A) top: Conventional 2D EPI where every slice is excited and acquired separately; middle: in multiplexed EPI several slices are excited simultaneously (colors indicate slice groups), allowing an acquisition speed-up given by the multiplexing factor; bottom: in 3D EPI the slice direction is replaced by a secondary phase encoding direction and the entire volume is continuously excited. With full sampling the speed is identical to 2D EPI, but parallel undersampling along k_z is now possible and allows substantial repetition time (TR) reductions. Panel (B) illustrates full (left) and factor-two accelerated (right) in-plane k-space sampling with 2D EPI. Panel (C) shows 3D EPI sampling schemes with factor 2 × 1 (left) and factor 2 along both primary (k_y) and secondary (k_z) phase encoding direction (right).

Figure 4

Example of a single 3D-EPI volume with 96 slices acquired in 2.3 s at a nominal resolution of 1 mm (A). A few slices have been selected and enlarged to be able to appreciate the fine anatomical details that are visible in a single functional image (B). The images have been bias field corrected using FMRIB’s Automated Segmentation Tool (FAST, FSL, [72] for visualization purposes. The complete acquisition parameters are: TR/TE/FA = 45 ms/17/ms/15deg, TR_vol = 2.34 s, matrix size 200 × 200, 96 slices + 9% slice oversampling, GRAPPA acceleration 4 × 2.