Spatio-temporal point-spread function of fMRI signal in human gray matter at 7 Tesla

Amir Shmuel, Essa Yacoub, Denis Chaimow, Nikos K. Logothetis, Kamil Ugurbil

Research output: Contribution to journalArticlepeer-review

172 Scopus citations

Abstract

This study investigated the spatio-temporal properties of blood-oxygenation level-dependent (BOLD) functional MRI (fMRI) signals in gray matter, excluding the confounding, inaccurate contributions of large blood vessels. We quantified the spatial specificity of the BOLD response, and we investigated whether this specificity varies as a function of time from stimulus onset. fMRI was performed at 7 Tesla (T), where mapping signals of parenchymal origin are easily detected. Two abutting visual stimuli were adjusted to elicit responses centered on a flat gray matter region in V1. fMRI signals were sampled at high-resolution orthogonal to the retinotopic boundary between the representations of the stimuli. Signals from macro-vessels were masked out. Principal component analysis revealed that the first component in space accounted for 96.2 ± 1.6% of the variance over time. The spatial profile of this time-invariant response was fitted with a model consisting of the convolution of a step function and a Gaussian point-spread-function (PSF). The mean full-width at half-maximal-height of the fitted PSF was 2.34 ± 0.20 mm. Based on simulations of confounding effects, we estimate that BOLD PSF in human gray matter is smaller than 2 mm. A time-point to time-point analysis revealed that the PSF obtained during the 3rd (1.52 mm) and 4th (1.99 mm) seconds of stimulation were narrower than the mean PSF obtained from the 5th second on (2.42 ± 0.15 mm). The position of the edge of the responding region was offset (1.72 ± 0.07 mm) from the boundary of the stimulated region, indicating a spatial non-linearity. Simulations showed that the effective contrast between active and non-active columns is reduced 25-fold when imaged using a PSF whose width is equal to the cycle of the imaged columnar organization. Thus, the PSF of the hyper-oxygenated BOLD response in human gray matter is narrower than that reported at 1.5 T, where macro-vessels dominate the mapping signals. The initial phase of this response is more spatially specific than later phases. Data acquisition methods that suppress macro-vascular signals should increase the spatial specificity of BOLD fMRI. The choice of optimal stimulus duration represents a trade-off between the spatial specificity and the overhead associated with short stimulus duration.

Original languageEnglish (US)
Pages (from-to)539-552
Number of pages14
JournalNeuroImage
Volume35
Issue number2
DOIs
StatePublished - Apr 1 2007

Bibliographical note

Funding Information:
This study was funded by NIH grants P41RR 08079, R01 MH 070800, and RO1 EB00331; The MIND Institute, W.M. Keck Foundation, and a European Molecular Biology Organization long-term fellowship awarded to AS. We thank Drs. Gregor Adriany and Peter Andersen for their excellent technical support, and Dr. Guenter Raddatz for comments on a previous version of the manuscript.

Keywords

  • BOLD
  • Blood oxygenation level dependent
  • Functional magnetic resonance imaging
  • Gradient echo
  • High-field MRI
  • High-resolution
  • Line spread function
  • Linear filters
  • Point-spread function
  • Retinotopic organization
  • Retinotopy
  • V1
  • Vision
  • Visual cortex
  • fMRI

Fingerprint Dive into the research topics of 'Spatio-temporal point-spread function of fMRI signal in human gray matter at 7 Tesla'. Together they form a unique fingerprint.

Cite this