Intravascular signal

Ideally, a perfusion image should be sensitive only to blood flow at the capillary level,

where exchange occurs. Signal from blood in large vessels, which are carrying blood

through the imaging slice to an exchange site in a more distal tissue region, is not desired.

This signal can be destroyed by the application o f bipolar gradients following the excitation

pulse (Ye et al., 1997). These gradient pulses dephase spins moving in the direction o f the

gradient, such that if a distribution o f velocities is present, the magnitude signal is

attenuated (a constant velocity within a voxel leads to a global phase change which has no

effect on magnitude images). However, the static tissue signal is refocused (apart from

diffusion effects) and therefore unaffected.

The application o f these bipolar gradients along all three gradient directions separately

would lead to the attenuation o f signal from blood moving in all directions. However, this

would extend the echo time considerably. Since the principal component o f vascular flow

within an axial imaging slice is expected to be along the slice-select direction, bipolar

gradients are commonly applied only along this direction in practice. Thus signal from

intravascular blood flowing into the slice with a component perpendicular to the imaging

slice is attenuated.

The 6-value (LeBihan et al., 1992) o f the bipolar gradients determines the level o f

attenuation o f the flowing signal. 6-values are calculated using the Stejskal-Tanner formula

(Stejskal and Tanner, 1965), modified to take the gradient ramps into account (since the

gradient pulses used in ASL are usually short, the contribution from the ramps cannot be

neglected):

6

[6.1]

where y is the gyromagnetic ratio, G is the gradient strength, and 5, A and 8 describe the gradient timings, as depicted in Fig. 6.9. The choice o f 6-value determines the minimum

velocity at which the signal is suppressed, which in turn is related to vessel diameter.

CHAPTER 6: FAIR IMPLEMENTATION 173

chosen such that vascular spins that flow straight through the voxel are crushed, whilst

those that irrigate the voxel contribute to the signal. However, this situation is not realistic;

the complexity o f the vascular network, which incorporates a range o f vessel diameters,

vessel orientations and blood velocities, means that the suppression o f blood in non­

exchanging vessels can never be completely effective.

Fig. 6.9: Diagram showing gradient timings for a bipolar crusher gradient.

Crusher gradients applied in the slice-select direction with 6-values between 0-6s/mm^

were tested in the sequence. This range covers typical values used in published ASL

studies. Higher 6-values were not tested in the interests o f keeping the TE short. Qualitative

inspection o f subtraction images acquired at 6-values ranging from 0-6s/mm^ in 6 healthy

volunteers showed a significant reduction in the prominence o f large vessels at 6=2s/mm^,

with no further improvement in image quality when using larger 6-values. At all 6-values

tested, some vessels remained visible, but since bipolar gradients were applied in only one

direction, elimination o f signal from all non-exchanging vessels was not expected. Figure

6.10 shows data from two o f the volunteers, which are representative o f all o f the

volunteers scanned. Two sets o f FAIR data collected from the same slice, one with

6=0s/mm^ (no intravascular crushers gradients), and one with 6=2s/mm^, are shown.

Sequence parameters were: Tl= 1200ms, 15 averages per TI, slice thickness 7mm,

labelling:imaging ratio 2.5:1, matrix size 64x64, FOV 250mm^, TD = 6s, TE = 25ms.

Visual comparison o f the FAIR images (which have been identically windowed for each

volunteer) shows a general signal decrease in image intensity due to the presence o f the

CHAPTER 6: FAIR IMPLEMENTATION 174

su b seq u en t F A IR im ages w ere ac q u ired w ith a 6 -v a lu e o f 2 s/m m (using G = 2 0 m T /m , 8= 0.6m s, ô = 4 .4 4 m s an d A =5.04m s).

m

A •

■. ' i ’M

Fig. 6.10: D ata from tw o h ea lth y v o lunteers. F L A S H an ato m ical im ages are show n in th e left colum n. T h e cen tral F A IR d ifferen ce im ages w ere acq u ired w ith no b ip o lar c ru sh e r g radients (6=0s/m m ^), and the rig h t-h an d im ages w ith 6=2s/m m ^, ap p lied in the slice-se lect direction. A ll F A IR d ata w ere acq u ired using T I= 1200m s. F o r each volu n teer, the tw o F A IR im ages h av e b ee n w in d o w ed identically; n o te th e general signal d ecrease in th e 6=2s/m m ^ im ages, p articu larly the sig n ifican t re d u c tio n in signal from large vessels.