Item

Abstract

Background:
The ability to reliably and reproducibly map high resolution functional architecture using fMRI techniques has
been a point of debate in animal as well as human studies. Several animal and human studies have successfully
mapped high resolution functional organizations, however, the robustness of the phenomenon (i.e. reproducibility
and demonstration in multiple subjects), which would certainly improve the credibility of the data, has been a
subject of debate, with only limited reproducibility demonstrated so far. Several different techniques have been
used (i.e. initial dip, blood flow, blood volume, Gradient Echo (GE) BOLD); however, which one is optimal
specifically for humans, has not been established. Here we demonstrate the spatial specificity of Hahn Spin Echo
(HSE) BOLD by robust mapping of ocular dominance columns in humans at 7 T. SNR and the BOLD effect both
increase with field strength, while the blood signal tends to be diminished because of the rapid shortening of its
T2. This bodes well for the accuracy of both GE and HSE BOLD. Previous studies in humans mapping high
resolution functional architecture used GE BOLD at 4T because the CNR of GE BOLD at 4T is much higher
compared to that of HSE BOLD. However, the success of GE BOLD relies on the expectation that non-specific
large vessel contributions will be equivalent for the two conditions and can be cancelled out. In the HSE
approach, this requirement is significantly alleviated. Furthermore, improved CNR and SNR at magnetic fields
exceeding 4T make the HSE BOLD technique feasible for such high resolution studies.
Methods and Results:
Studies were conducted at 7 T using slab selective FOV reduction for HSE (TR/TE 6000/50 ms) and 3 image
segments. Resolution (60 x 256): 0.5 x 0.5 x 3 mm3. To suppress gross subject motion, a bite bar was used. In
addition, any scans with significant motion were discarded. The visual stimuli were presented through fiber optic
video goggles. Imaging was performed on a region of flat cortex where the ODCs are expected to run
perpendicular to the inter-hemispheric fissure. Subjects were brought back for several repeated studies to assess
the reproducibility of the functional maps. Approximately the same anatomical location was selected in multiple
sessions with the same subject. The maps from different sessions were co-registered to allow for (see Figure)
identification of the same columns in the different sessions. The Figure presents an ODC map from one subject
obtained in three different sessions (over the period of several months). Despite inevitable but small differences in
the slice positioning, highly reproducible maps were obtained. Similar reproducibility was seen in 3 different
subjects.
Conclusions:
Using HSE BOLD at very high magnetic fields (7T), we have demonstrated not only that high field fMRI can
produce reliable and reproducible maps at the sub-millimeter level but also that these maps have intrinsically high
spatial specificity, even to the level of cortical columns in humans. We expect HSE BOLD to be the imaging
choice for high resolution applications in humans at high fields.