Quantitative Arterial Spin Labelling ASL at UltraHigh Field 7.0 T

1
Quantitative Arterial Spin Labelling (ASL) at
Ultra-High Field (7.0 T) A. G. Gardener, P. A.
Gowland and S. T. Francis Sir Peter Mansfield
Magnetic Resonance Centre, School of Physics and
Astronomy, University of Nottingham, University
Park, Nottingham, NG7 2RD, UK.

• Introduction and Theory
• At ultra-high field arterial spin labelling
  (ASL) using magnetically-labelled blood as an
  endogenous tracer - benefits from an overall
  increase in SNR and T1 relaxation times of blood
  and tissue. This leads to increased perfusion
  weighted (PW) signals occurring at longer delay
  times (TIs).
• Here two Pulsed ASL schemes, STAR (Signal
  Targeting with Alternating Radiofrequency) 1
  and FAIR (Flow sensitive Alternating Inversion
  Recovery) 2 are assessed in humans at 7.0 T.
• STAR and FAIR PASL techniques are based on the
  subtraction of two consecutively acquired label
  and non-label images. STAR (Fig. 1A) uses a label
  inversion slab (tag) applied proximal and a
  non-label slab (control) applied distal to the
  imaging slab. FAIR (Fig. 1B) uses a selective
  inversion slab for the label, and a non-selective
  inversion pulse for the non-labelled image,
  centred over the imaging slab. In both cases the
  difference between the label and non-label images
  provides the PW signal.
• However B1 and B0 inhomogeneity at 7.0 T can
  lead to imperfect inversion pulse profiles,
  resulting in offsets between label and non-label
  images in the absence of perfusion.
• One solution is to increase the gap between the
  inversion slab and image slice, by altering the
  width of the label/non-label, and/or by applying
  pre-/post-saturation pulses to the imaging slices
  immediately before/after the inversion pulses.
  However, these can all have an adverse effect on
  transit time.
• These factors are assessed to determine the
  optimal PASL sequence for use at 7.0 T.

Figure 1 (A) STAR and (B) FAIR ASL
Schemes.

• Method and Analysis
• Data were acquired on a 7.0 T Philips Achieva
  scanner. For inversion a hyperbolic secant (hsc)
  adiabatic pulse was used, whilst for
  pre/post-saturation an optimized sinc-gauss pulse
  was applied. Initial experiments performed on
  phantoms assessed inversion pulse efficiency, a,
  as a function of position for various slab
  widths/offsets. Profiles of labelling slabs were
  measured transaxially using sagittal SE-EPI
  images.
• For the ASL sequences, an image slab of five
  GE-EPI slices with in-plane resolution of 3 x 3
  mm2, 3 mm slice thickness with 1 mm slice gap, 64
  x 64 matrix and echo time of 20 ms were acquired
  in ascending order after each labelling scheme.
• STAR used 70 mm wide label and non-label slabs
  label gaps of 10-25 mm between image/inversion
  slabs. FAIR used selective label slabs of 30 -
  50 mm and 150 mm non-selective non-label slabs.
  Widths of labels/non-labels were chosen to allow
  for full refreshment of blood in the TR period.
  The effect of the addition of pre-/post-saturation
  pulses immediately before and after the
  inversion pulse (up to 2 of each) and different
  crusher gradient areas after each saturation
  pulse was assessed.
• The optimal STAR and FAIR labelling schemes were
  performed on 4 healthy human volunteers, who gave
  informed written consent. Images were acquired at
  five TIs, with sixty ASL sets acquired at each
  TI, with a TR of 3 s. Images were acquired both
  with and without velocity crushing (diffusion
  weighting 2.5 smm-1, critical velocity 20 mms-1).
  A T1 map and M0 map were also generated for
  perfusion quantification. PW images were formed
  from (label - non-label) images and
  quantification of perfusion performed using a
  multi-compartment system model, assuming the T1
  of arterial blood to be 2.0 s.

Figure 3 (A) STAR vs. (B) FAIR at TI
800 ms. For these scans no vascular crushing was
applied.

• Results and Discussion
• Figure 2 shows inversion efficiency as a
  function of label width for a label positioned at
  coil isocentre and also off-centre by label
  width. The inversion efficiency is gt0.98 over the
  central 10 cm of the label, but profile sharpness
  is poor and varies depending on position in the
  coil. The hsc pulse had a simulated inversion
  profile Mz lt -0.98 M0 for 86 of slice width,
  which gave significant static tissue artefact for
  STAR in the difference images, but not for FAIR.
• The effect of using pre- and post-saturation
  pulses to suppress the offset signal for STAR
  label gap of 25 mm was assessed. It was found
  that the application of a single pre- and post-
  saturation pulse with variable crushers gave the
  optimal reduction of offset signal (0.15 ). The
  same scheme reduced the offset signal to 0.08
  in FAIR and this saturation scheme was used in
  subsequent human subject studies. STAR labels
  were set to be 70 mm wide, whereas FAIR could use
  a 50 mm selective 15 mm effective label gap and
  a 150 mm non-selective slab, resulting in shorter
  transit time delays for FAIR.
• Figure 3 shows averaged difference images at a
  TI of 800 ms for the two schemes, showing the
  unwanted static tissue signal artefact in the
  STAR PW image set. Figure 4 shows a perfusion map
  formed from fitting five increasing TI FAIR
  data-sets. Figure 5 shows corresponding signal
  change with TI in two grey matter ROIs with
  vascular crushing applied for FAIR.

• Conclusions
• At 7.0 T STAR gives large artefactual signal
  from static tissue, due to poor label slab
  profiles caused by B1 inhomogeneity. This problem
  was worse away from the isocentre of the coil.
  The STAR static tissue artefact could not be
  fully suppressed by saturation pulses even at
  wider label gaps, which led to longer long
  transit times and so reduced label efficiency.
• The FAIR labelling scheme is less sensitive to
  offset effects as the image slab is positioned at
  the centre of the labels, and hence the label
  centres are generally in the homogeneous region
  of the coil. With pre- and post-saturation pulses
  good static tissue suppression could be obtained
  for FAIR at small label gaps.
• At 7.0 T quantitative measures of perfusion are
  best obtained with FAIR using in-plane
  saturation, with PW signal intensity in grey
  matter of 1.5 and signal change peaking at TIs
  of 1800 ms.

References and Acknowledgments 1) Edelman et al,
MRM, 40, 513 (1994). 2) Kim et al, MRM, 34,
293-301 (1995). 3) Pfeuffer et al, MRM, 47, 903
(2002). Funded by the MRC and EPSRC.