Christopher C. Conlin^{1} and Jeff L. Zhang^{1}

This study outlines an approach for selecting optimal TIs at which to sample renal ASL data. We present an error-propagation factor for a model of the ASL signal and propose to optimize TI sampling through minimization of this factor. Using FAIR ASL data from 7 human subjects, we show that renal perfusion estimates obtained with optimal TI sampling are more accurate and precise than estimates obtained with uniform TI sampling, particularly when ASL data is acquired at only a few TIs.

In this IRB-approved study, ASL data were acquired from 7 healthy volunteers at 3T (TimTrio; Siemens) using FAIR tagging and bSSFP readout: TR 3.68ms, TE 1.84ms, FOV 380×380mm, matrix 256×256, slice thickness 8mm, centric reordering, and GRAPPA factor 2. For each patient, data were acquired at 16 TIs: 150ms, then 200ms–1600ms at 100ms intervals. At each TI, SS and NS images were acquired at end-inspiration from a coronal slice through the kidneys (Figure 1).

SS and NS signal-versus-TI curves were obtained for each kidney by averaging the signal intensity within cortical and medullary ROIs on the SS and NS images at each TI. Subtraction of NS signals from SS signals yielded ASL difference-signal curves for each kidney. Reference perfusion measurements were obtained by fitting these 16-TI difference-signal curves with equation 1.

The 16-TI difference-signal
curves were then downsampled to produce curves with 4 TIs and 8 TIs. The TIs
included in the 4-TI and 8-TI curves were chosen from the original 16 using two
strategies: 1) Uniform sampling, wherein TIs were uniformly distributed along
the original curve, and 2) Optimized sampling, wherein TIs were determined by
minimizing equation 3 and selecting TIs from the original 16 that most closely
matched the optimal TIs from the minimization. Equation 3 was minimized over expected
renal-parameter ranges: F: 150–650mL/min/100g, t_{0}: 0–200ms, and T_{1}:
800–2000ms.

Perfusion was
estimated from the downsampled curves by fitting with equation 1. The accuracy of perfusion estimates from the downsampled curves was evaluated by computing the absolute
difference between reference perfusion estimates and those from the downsampled curves
(ΔF = |F_{reference} – F_{downsampled}|). Correlation between reference
perfusion values and those from downsampled curves was also determined.

1. Alsop DC, Detre JA, Golay X, et al. Recommended implementation of arterial spin-labeled perfusion MRI for clinical applications: A consensus of the ISMRM perfusion study group and the European consortium for ASL in dementia. Magn Reson Med 2015;73(1):102-116.

2. Conlin CC, Zhao Y, Zhang JL. Improving the accuracy of renal perfusion measurements from ASL by using multiple TIs: Validation with DCE MRI. Proc Int Soc Magn Reson Med 24 (2016).

3. Zhang JL, Sigmund EE, Rusinek H, et al. Optimization of b-value sampling for diffusion-weighted imaging of the kidney. Magn Reson Med 2012;67(1):89-97.

4. Zhang JL, Koh TS. On the selection of optimal flip angles for T1 mapping of breast tumors with dynamic contrast-enhanced magnetic resonance imaging. IEEE Trans Biomed Eng 2006;53(6):1209-1214.