Single-shot half k-space high-resolution gradient-recalled EPI for fMRI at 3 Tesla

Magn Reson Med. 1998 Nov;40(5):754-62. doi: 10.1002/mrm.1910400517.

Abstract

Half k-space gradient-recalled echo-planar imaging (GR-EPI) is discussed in detail. T2* decay during full k-space GR-EPI gives rise to unequal weighting of the lines of k-space, loss of signal intensity at the center of k-space, and a point-spread function that limits resolution. In addition, the long readout time for high-resolution full k-space acquisition gives rise to severe susceptibility effects. These problems are substantially reduced by acquiring only half of k-space and filling the empty side by Hermitian conjugate formation. Details of the pulse sequence and image reconstruction are presented. The point-spread function is 3(1/2) times narrower for half than full k-space acquisition. Experiments as well as theoretical considerations were carried out in a context of fMRI using a whole-brain local gradient and an RF coil at 3 Tesla. Using a bandwidth of +/-83 kHz, well-resolved single-shot images of the human brain, as well as good quality fMRI data sets were obtained with a matrix of 192 x 192 over 16 x 16 cm2 FOV using half k-space techniques. The combination of high spatial resolution using the methods presented in this article and the high temporal resolution of EPI opens opportunities for research into fMRI contrast mechanisms. Increase of percent signal change as the resolution increases is attributed to reduction of partial volume effects of activated voxels. Histograms of fMRI pixel responses are progressively weighted to higher percent signal changes as the resolution increases. The conclusion has been reached that half k-space GR-EPI is generally superior to full k-space GR-EPI and should be used even for low-resolution (64 x 64) EPI.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Brain / anatomy & histology*
  • Echo-Planar Imaging / methods*
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Models, Neurological
  • Phantoms, Imaging
  • Sensitivity and Specificity