Neuroimaging Assessment of Subarachnoid Hemorrhage - Animal Models of Acute Neurological Injuries: Injury and Mechanistic Assessments - page 432

based on the severity of the artifact you are trying to remove;
a larger fi lter will remove more artifacts but will also blur and
degrade the image. Theoretically, the smallest fi lter size that
results in a usable image is what you want to use. However, to
compare two fi ltered phase images they must be fi ltered with
the same size of fi lter (as the fi lter affects the phase values) so
choosing a larger fi lter to allow for some variance in collected
image quality is advisable. Recommended fi lter sizes are from
32 to 64 depending on the severity of artifacts.
7. The SW images are a combination of the fi ltered phase and the
acquired magnitude. The contrast in magnitude image is
enhanced by darkening regions of positive phase, negative
phase, or both (any phase changes). Typically, regions of
increased fi eld strength, or positive susceptibility, are chosen to
be enhanced. Unfortunately, this can correspond to either pos-
itive phase (in a right-handed system) or negative phase (in a
left-handed system). This depends solely on the convention of
the manufacturer of the system, for example Bruker is a left-
handed system and Siemens is a right-handed system. An easy
way to check this is to look at the phase of the veins. If the
veins appear dark, it is a left-handed system; if the veins appear
bright, it is a right-handed system (as the veins have positive
susceptibility). For a left-handed system that shows dark veins
normal SWI convention is to darken areas of negative phase
(use negative phase masking).
8. To generate the SW images, a phase mask is created where
negative (for negative masking) phase values are linearly
mapped from 0 to 1 and positive phase values are set to 1. This
mask is then multiplied by the magnitude image n times. This
will selectively darken areas with negative phase while areas
with 0 or positive phase will remain unaffected. The value of n
can be adjusted to increase or decrease the amount that regions
of negative phase are darkened. In general, a value of n = 4 was
found to have optimal contrast-to-noise for a wide variety of
phase values ( 37 ) and should usually be used.
9. Minimum intensity projections of the SW images can be gen-
erated to improve visualization of veins. An mIP collapses mul-
tiple images into a single image by taking the minimum value
at each location ( x , y ). This is most useful for visualizing venous
connectivity over multiple slices. Generating an mIP across too
many slices will lose the edges of the brain as the black back-
ground from the smallest slice will cover the edges of larger
slices. The noise in an mIP also increase for larger numbers of
slices. In human data, an mIP over about 8 mm is generally
good, while in animal data <1 mm (usually 2-3 slices) is
preferred.