Diffusion weighted imaging of hyperpolarized (HP) substrates has the potential to identify both the production and microstructural compartmentalization of HP metabolites, providing unique insight into tumor metabolism and lactate export. However, any deviation from the expected signal evolution will lead to modeling bias and parameter mis-estimation. Slice profile effects, which have previously been observed in HP gas experiments¹, are caused by an imperfect excitation profile that results in through-slice flip angle variation. This yields excess signal in the slice transition region after repeated RF pulses, resulting in k-space filtering for a Cartesian acquisition. The single-shot imaging techniques used in HP ¹³C imaging are especially sensitive to this effect, as the excess signal will directly alter the signal intensity in each diffusion weighted image. Here, slice profile effects can significantly bias the apparent diffusion coefficient (ADC) measurements of HP substrates, where only a few images (or spectra) are acquired at different b-values and high flip-angles^2,3,4. The goal of this study was to determine the impact of slice profile effects on ADC mapping of HP substrates. Digital simulations as well as in vivo ADC mapping using HP ¹³C urea in the murine liver were performed.

HP spin simulations were performed using the RF pulse design library⁵ in MATLAB. The slice profile was simulated for a Gaussian pulse shape, and the signal for each flip-angle was determined by integration along the slice direction. Three flip angle schedules were simulated (Fig. 1): a constant 30° flip angle, an RF compensated flip schedule (RF VFA)⁶, and an RF and diffusion compensated flip schedule (RF, DWI VFA)³. To account for deviation due to slice profile effects, the slice-select gradient was scaled to yield the desired signal response. To assess these effects in vivo, 90mg of [¹³C,¹⁵N₂]urea was polarized in a 3.35T HyperSense for 90 minutes and rapidly dissolved with PBS and 0.3mM EDTA. 350μL of 80 mM urea was injected via tail-vein over 15s. Diffusion weighted imaging was performed using a Varian 14T MRI scanner 30s after the start of injection using a double spin-echo sequence and a single shot, flyback EPI readout, with a variable flip-angle schedule to account for both RF decay and diffusion weighting³. Respiratory gating was used to minimize aberrant signal loss from bulk motion. The slice-select gradient was either held constant or scaled to account for the simulated deviation due to slice profile effects. Eight b-values were acquired in decreasing value, ranging from 700 to 50 s/mm² in a total of 1-2s. Raw data were corrected for prior RF pulses and subsequently fit voxelwise to a monoexponential decay:

$$S_{n,corr} = \frac{S_{n,raw}}{\sin\theta_{n}\times\prod_{k=1}^{n-1}\cos\theta_{k}}$$
$$S(b)=S_{0,corr}\exp(-bD)$$

Simulations of the slice profile for HP spins can be seen in Figure 2. While the shape of the excitation profile is constant throughout the experiment, the longitudinal magnetization does not recover between TRs for hyperpolarized spins and is therefore depleted non-uniformly by previous excitations. This results in a distorted slice profile after repeated RF pulses, giving rise to an effectively wider slice and excess signal (Fig. 2A,D,G). This excess signal can be corrected for by scaling the slice-select gradient, minimizing excitation of the transition band while conforming to the desired signal response (Fig. 2B,E,H). These slice profile effects accumulate after repeated RF pulses, leading to maximum excess signal of 1.65 (30°, Fig. 2C), 2.0 (RF VFA, Fig. 2F) and 2.7 (RF, DWI VFA, Fig. 2I) for the three flip angle schemes with a Gaussian pulse shape. This, in turn, results in a systematic bias when b-values are acquired sequentially, leading to overestimation of the measured ADC (Fig. 3). The greatest bias is seen for small ADCs, resulting in 5 to 6-fold overestimation for metabolites in more restricted environments. In vivo, the murine liver ADC value of ¹³C urea (Fig. 4) without accounting for slice profile effects is, on average, nearly a factor of two greater compared to applying the slice-select gradient correction (1.38 ± 0.15 vs 0.70 ± 0.11 x 10^-3 mm²/s). The magnitude of this overestimation is in good agreement with the estimated bias from digital simulations seen in Fig. 3.Hyperpolarized diffusion weighted imaging has the potential to provide non-invasive measures of transporter function and metabolite microenvironments. However, the imperfect slice profile and the transient, non-recoverable hyperpolarization lead to excess signal at later flip-angles, potentially biasing ADC estimation. Gradient scaling can correct for this deviation, minimizing bias and providing more precise measurements for quantitative imaging.