08-05 Faster Image Acquisition by k-Space Manipulation
A completely different approach to accelerate image acquisition is through faster acquisition of image data rather than optimizing pulse sequences. A number of different approaches have been proposed:
Reduced Acquisition. Instead of acquiring, for instance, 256 lines, we acquire only 80% and zero-fill the remaining lines. We loose some of the spatial resolution, but for many clinical applications, the raw data are sufficient (Figure 08- 09a; cf. Figure 07-07).
Halfscan. In this case we acquire an asymmetrical fraction of the data set. The rest of the data are replaced by the symmetrical data from the other side of k-space. Spatial resolution is maintained, but there is a loss of signal-to-noise (Figure 08-09b).
![]() |
![]() |
Figure 08-09: |
Rectangular Field-of-View. The final MR image can be turned into a rectangular image by collecting only half the lines in k-space. By doing this, image time, as well as the field-of-view, will be halved, which is convenient for imaging of the extremities and the spine, or in angiography (Figure 08-10). However, the signal-to-noise ratio will also be substantially reduced.
![]() |
![]() |
Figure 08-10: |
k-Space Substitution. To accelerate dynamic image data acquisition, one can apply k-space substitution, also called ‘keyhole’ imaging [⇒ Jones; ⇒ van Vaals]. This technique collects the entire k-space of a reference image; for the subsequent images, however, only the central lines are recorded. These data are then combined with the outer lines of the reference data space to add information on edge definition and sharpness. In this way, the uptake of a contrast agent can be followed very rapidly (Figure 08-11).
![]() |
![]() |
Figure 08-11: |
Spiral (Helical) and Radial Scanning. Alternatives to filling k-space line-by- line are spiral (helical) or radial scanning techniques (Figure 08-12). These methods are very fast and therefore suited for dynamic imaging and, e.g., for cardiac imaging. They use projection reconstruction (backprojection) algorithms as described earlier.
![]() |
![]() |
Figure 08-12: |