Biomedical Engineering Reference
In-Depth Information
FIGURE 5.1
Illustration of geometric distortion due to gradient non linearity in MRI. On the left, the
automatic correction scheme (“Gradwarp,” see reference 8 in Moerland et al.
25
) has not
worked properly following the selection of a rectangular field of view (24 cm
18 cm).
This is reflected in the apparent displacement of the middle fiducial markers of a stereo-
tactic frame away from the imaginary lines joining the superior and inferior fiducials; the
red lines have been superimposed to emphasize this point. The distortion was so severe
as to render the surgical procedure unsafe. On the right, the distortion is significantly
reduced following the selection of a square field of view (24 cm
24 cm). See reference
26 for more details.
described as “scanner error,” its effect is physically indistinguishable from
the others and is present in virtually all images. Furthermore, the presence of
metalic objects (stereotactic frame, surgical instruments, surgical implants,
etc.) within the field of view will generally result in severe distortions. The
geometric distortions due to susceptibility differences can be extremely
complex due to field inhomogeneity dependency on both the material and
shape of the imaged object, as well as its orientation relative to the static
field. On the other hand, distortions resulting from magnet imperfections
and eddy currents can often be modeled. Field inhomogeneity causes local
variations in resonance frequency and therefore is sometimes referred to as
an off-resonance effect. The consequence of this is that the resulting distor-
tion is inversely proportional to gradient strength (bandwidth per pixel)
and therefore can be minimized by proper selection of gradient strength
and field of view. In standard spin warp imaging, the effect is limited to the
frequency encode direction. In echo-planar imaging (EPI), the effect is
present in both the read and phase encode (“blip”) directions, the latter
dominating because the bandwidth per pixel is lowest in the phase encode
direction.
Geometric distortion in the phase encode direction in EPI can
be very large and is still an important problem in many applications, such
as diffusion-weighted imaging (DWI) and functional MRI; see Figure 5.2
for an illustration of the distortions in EPI. Another important source of
distortions in EPI-DWI is eddy currents,
27
28
which are particularly problem-
atic because the size of the distortion is a function of the amplitude of the
diffusion-sensitizing gradients; therefore, it can vary during the acquisi-
tion process.
