Demonstration of magnetization transfer and relaxation normalized pH-specific pulse-amide proton transfer imaging in an animal model of acute stroke

Abstract

Purpose:

pH-weighted amide proton transfer (APT) MRI is promising to serve as a new surrogate metabolic imaging biomarker for refined ischemic tissue demarcation. APT MRI with pulse-RF irradiation (pulse-APT) is an alternative to the routine continuous-wave (CW-) APT MRI that overcomes the RF duty cycle limit. Our study aimed to generalize the recently developed pH-specific magnetization transfer and relaxation-normalized APT (MRAPT) analysis to pulse-APT MRI in acute stroke imaging.

Methods:

Multi-parametric MRI, including CW- and pulse-APT MRI scans, were performed following middle cerebral artery occlusion in rats. We calculated pH-sensitive MTR_asym and pH-specific MRAPT contrast between the ipsilateral diffusion lesion and contralateral normal area.

Results:

An inversion pulse of 10 to 15 ms maximizes the pH-sensitive MRI contrast for pulse-APT MRI. The contrast to noise ratio of pH-specific MRAPT effect between the contralateral normal area and ischemic lesion from both methods are comparable (3.25 ± 0.65 vs. 3.59 ± 0.40, P>0.05). pH determined from both methods were in good agreement, with their difference within 0.1.

Conclusions:

Pulse-APT MRI provides highly pH specific mapping for acute stroke imaging.

Introduction

The ischemic tissue transitions from aerobic to anaerobic metabolism upon acute ischemia, resulting in pH drop, and therefore, pH-sensitive MRI is promising to serve as a surrogate metabolic biomarker of acute stroke (1,2). Amide proton transfer (APT), a form of chemical exchange saturation transfer (CEST) MRI, is pH sensitive (3,4). pH MRI has been translated to study acute stroke, both in animal models (5–8) and patients (9–12). pH imaging enables classification of the perfusion/diffusion lesion mismatch into perfusion/pH lesion mismatch (benign oligemia) and pH/diffusion lesion mismatch (metabolic penumbra), refining the routine mismatch paradigm (13,14). Indeed, recent clinical trials have shown that response to endovascular reperfusion is not time-dependent in patients with salvageable tissue, underscoring the importance of accurate penumbra mapping (15–17). Although CEST MRI has often been implemented with continuous wave (CW) RF irradiation, the irradiation scheme of an RF pulse train is of reduced RF duty cycle and specific absorption rate (SAR), making it amenable to all scanners (18). For slow exchangeable protons, it has been shown that pulse-CEST MRI with inversion pulses of intermediate durations provides similar contrast to that of CW-CEST MRI (19–21).

The routine magnetization transfer (MT) ratio asymmetry (MTR_asym) analysis, despite its pH sensitivity, is susceptible to the slightly asymmetric magnetization transfer (MT) effects. MTR_asym is heterogeneous across the intact white and gray matter despite their little pH difference (22). It has been shown that in rats, the T_1w-normalized CEST effects for the intact brain WM and GM are 2.08±0.20 vs. 2.00±0.23%/s (APT), −1.79±0.88 vs. −0.89±0.97%/s (MT contrast asymmetry) and 4.42±0.58 vs. 3.38±0.42%/s (nuclear overhauser enhancement, NOE), respectively (23). Indeed, the noticeable NOE and MT contrast between brain WM and GM despite their little pH difference suggest that NOE and MT effects are not pH-specific. The recently developed MT and relaxation-normalized APT (MRAPT) analysis minimizes the intrinsic non-pH related heterogeneity in pH-weighted MTR_asym image toward pH-specific imaging (24,25). Although it has been shown that B₀ inhomogeneity can be corrected in phantoms using an analytical approach that accounts for the direct saturation effect (26), the water saturation shift referencing (WASSR) approach has been commonly used for correcting field inhomogeneity (27). We have recently demonstrated that the MRAPT analysis enables expedient correction of moderate field inhomogeneity without the Z spectrum (28). Our study here aimed to extend pH-specific MRAPT analysis to pulse-CEST at 3.5 ppm (i.e., pulse-APT) for acute stroke imaging. Towards this goal, we performed both CW- and pulse-APT MRI in a rodent model of acute stroke. We evaluated the effects of RF saturation duration and flip angle on pH-weighted contrast and confirmed the optimal experimental condition for the pulse-APT imaging. Capitalizing on the highly correlated MRAPT analysis of CW- and pulse-APT MRI measurements, we demonstrated that tissue pH could be reasonably derived from pulse-APT MRI.

Theory

The in vivo MTR_asym can be described by a summation of pH-sensitive APT signal and concomitant RF irradiation effect (8). We have

$${\text{MTR}}_{\text{asym}}=\frac{{\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}}{{\text{R}}_{1\text{w}}}·\text{α}·\left(1-\text{σ}\right)+\text{ϵ}\left({\text{MTR}}_{\text{asym}}^{\prime }\right)$$ (1)

where f_s and k_sw are the labile amide proton ratio, and exchange rate, respectively, and α and σ are the effective labeling coefficient and spillover factor (29). The term $\mathrm{ϵ}\left({\mathrm{M}\mathrm{T}\mathrm{R}}_{\mathrm{a}\mathrm{s}\mathrm{y}\mathrm{m}}^{\prime }\right)$ denotes the non-APT baseline shift (30). It has been shown that for the condition of weak RF irradiation (e.g., 0.75 μT at 4.7 T MRI), such a baseline shift is not pH-sensitive (31–34). It has been shown that the labeling coefficient and spillover factor at 3.5 ppm is nearly uniform and approximately 1 with a moderate RF irradiation field (8).

Because the intact tissue has little pH difference (22), we have shown that the regression of R_1w·MTR_asym between the normal white and gray matter accounts for the non-pH background variation (i.e., R_1w·MTR_asym = F(R_1w, MMTR)), in which MMTR is the mean MTR at +3.5 and −3.5 ppm. For the ischemic tissue, we have

$$\begin{gathered} {\left({\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}\right)}_{ischemic}={\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}-\text{ϵ}\left({\text{MTR}}_{\text{asym}}^{\prime }\right)\right)}_{ischemic} \\ \approx {\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}\right)}_{ischemic}-{\text{R}}_{1\text{w}}·\text{ϵ}{\left({\text{MTR}}_{\text{asym}}^{\prime }\right)}_{norm} \\ \approx {\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}\right)}_{ischemic}-{\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}-\frac{{\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}}{{\text{R}}_{1\text{w}}}\right)}_{norm} \\ \approx {\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}\right)}_{ischemic}-{\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}\right)}_{norm}+{\left({\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}\right)}_{norm} \end{gathered}$$ (2)

Because it has been shown that semisolid MT and NOE effects are not highly pH-sensitive under weak RF saturation, the pH specific change in MRAPT ratio (i.e., ΔMRAPTR) is directly related to the ischemia-induced exchange rate drop as,

$$\begin{gathered} \text{ΔMRAPTR}={\left({\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}\right)}_{ischemic}-{\left({\text{f}}_{\text{s}}·{\text{k}}_{\text{sw}}\right)}_{norm} \\ \approx {\left({\text{R}}_{1\text{w}}·{\text{MTR}}_{\text{asym}}\right)}_{ischemic}-{\text{R}}_{1\text{w}}·{\left({\text{MTR}}_{\text{asym}}\right)}_{norm} \\ \approx {\left({\text{R}}_{1\text{w}}·{\text{MTR}}_{\text{asym}}\right)}_{ischemic}-\text{F}{\left({\text{R}}_{1\text{w}},\text{MMTR}\right)}_{norm} \end{gathered}$$ (3)

where (f_s·k_swischemic − (f_s·k_sw)_norm is pH-sensitive APT MRI contrast. For acute stroke imaging, it is generally believed that pH dominates the APT effect (i.e., (f_s·k_sw)_ischemic − (f_s·k_sw)_norm = f_s · Δk_sw). Because MRAPT analysis accounts for concomitant non-pH sensitive heterogeneity, it simplifies multi-pool (i.e., APT, NOE and MT, etc.) into a simple 2-pool model (i.e., APT) and becomes more specific to tissue pH change than MTR_asym (24).

Materials and methods

Animals

The animal experiments have been approved by the Institutional Animal Care and Use Committee, Massachusetts General Hospital. We imaged five adult male Wistar acute stroke rats (N=5). Anesthesia was induced with 5% isoflurane/air mixture for 5 minutes and then maintained under 1.5–2% isoflurane throughout the study. We chose a classical middle cerebral artery occlusion (MCAO) model by gently inserting a 4–0 silicone-coated nylon suture through the internal carotid artery (ICA) to block the origin of the middle cerebral artery (MCA). Rats were imaged about 2 hours after the MCAO operation. Animal heart rate and blood oxygen saturation level were monitored (Nonin Pulse Oximeter 8600, Plymouth, MN) with their core temperature maintained by a circulating warm water jacket surrounding the torso.

MRI

All MRI scans were performed using a 4.7 Tesla small-bore MRI system (Bruker Biospec, Ettlingen, Germany). We used single-shot echo planar imaging (EPI) with a field of view (FOV) of 25×25 mm² (image matrix = 64 × 64, slice thickness/slice gap = 1.8/0.2 mm, 5 slices with a bandwidth of 200 kHz). Multi-parametric MRI scans were performed, including perfusion, diffusion, T₁, T₂, and CEST MRI. Briefly, T₁–weighted inversion recovery images were acquired with inversion delays ranging from 250 to 3000 ms (repetition time (TR)/echo time (TE)=6500/20 ms, 4 averages, scan time= 3 min 38 s), and T₂–weighted MRI was obtained with two spin echo EPI scans (TR/TE1/TE2=3250/30/100 ms, 16 averages, scan time= 1 min 40 s). The amplitude-modulated arterial spin labeling (AM-ASL) perfusion MRI was performed (TR/TE=5000/20 ms, ASL tagging time=3000 ms, 32 averages, scan time=5 min 36 s) (35). Diffusion MRI was obtained with a single-shot isotropic diffusion-weighted MRI (two b-values of 250 and 1000 s/mm², TR/TE=3250/42 ms, 16 averages, scan time =1 min 44 s)(36). We performed the fast multislice CEST MRI with an unevenly segmented RF irradiation scheme with gradient-echo readout to acquire CEST images (37). We obtained a Z-spectrum with CW-CEST MRI, from −6 to 6 ppm with intervals of 0.25 ppm (B₁=0.75 μT, relaxation time=3500 ms, primary saturation time=3000 ms, secondary saturation time = 500 ms, TE= 15 ms, 4 averages, scan time = 13 min 20s). We used three-point (±3.5 ppm, B₁=0.75 μT) CW- and pulse-APT MRI (Gaussian-shaped RF pulse train (FA=180°) of a 50% RF duty cycle with spoiler gradients between RF pulses), with 8 averages of the control scan and 32 averages of saturated scans at ±3.5 ppm (scan time = 4 min) in 5 acute stroke rats. The RF pulse duration and flip angle (FA) were varied for the optimization of pulse-APT MRI. In one rat, we fixed the RF duration to 15 ms and varied the FA from 60° to 540° with intervals of 60°. In another rat, we serially varied the RF pulse duration (i.e., τ) from 5 to 30 ms with intervals of 5 ms, for an FA of 180°.

Data Analysis

Images were processed in Matlab R2019a (Mathworks, Natick, MA). Parametric apparent diffusion coefficient (ADC), and T₁ and T₂ maps were obtained using least-squares mono-exponential fitting of the signal intensities as a function of the diffusion b-value (I₀e^−b·ADC), inversion time (I = I₀|1 − (1 − η)·e^−TI/T₁|, where η is the inversion efficiency), and echo time (I = I₀e^−TE/T₂), respectively. The cerebral blood flow (CBF) was calculated using ($CBF=\frac{\lambda }{T_{1app}}·\frac{I_{ref}-I_{tag}}{2\alpha ·I_{ref}}$ , where λ is the brain/blood partition coefficient, I_tag and I_ref are perfusion tagging and reference images, respectively, α is the degree of blood spin inversion, and T_1app is the longitudinal relaxation time in the presence of RF labeling pulse (35). The B₀ field inhomogeneity was determined using the WASSR approach, based on the CW-CEST Z-spectrum. The mean MTR (MMTR) and MTR_asym was calculated as $1-\frac{I_{ref}+I_{label}}{2I_0}$ and $\frac{I_{ref}-I_{label}}{I_0}$, respectively, where I_ref and I_label are the reference (−3.5 ppm) and label (+3.5 ppm) scans while I₀ is the control image without RF saturation. The ischemic diffusion lesion was automatically segmented from the ADC map using the K-means clustering algorithm (38). We calculated the contrast to noise ratio (CNR) as the ratio of the difference and standard deviation from the ischemic lesion and the contralateral normal tissue. A paired t-test was performed to evaluate the ischemia-induced MRI changes, and P values less than 0.05 were regarded as statistically significant.

Results

Fig. 1 shows parametric T_1w, T_2w, and ADC maps from a representative acute stroke rat. Table 1 summarizes multi-parametric signals from the ipsilateral ischemic lesion and contralateral normal area and their difference, from all stroke rats. T_1w (Fig. 1a) increased from 1.54 ± 0.08 (contralateral normal tissue) to 1.68 ± 0.05 (ischemic lesion, P<0.05). T_2w change was small and insignificant (from 54.81 ± 1.84 to 54.50 ± 1.53 ms, Fig. 1b). The diffusion lesion was segmented using a K-means clustering algorithm and mirrored to the contralateral normal brain. The diffusion lesion volume was 122±96 mm³ (mean ± std). The ischemic lesion had significant ADC drop (Fig. 1c) from the normal tissue (0.63 ± 0.05 vs. 0.86± 0.02 μm²/ms, P<0.001). Fig. 1d shows the pH-sensitive MTR_asym image obtained from a three-point CW-APT MRI. The hypointensity in the ischemic lesion (−5.78 ± 0.38 % vs. −3.63 ± 0.63 %, P<0.01) is a hallmark of tissue acidosis following anaerobic glycolysis. Z-spectra and the corresponding asymmetry spectra from CW-CEST (Supporting Information Figures S1a and S1b) and pulse-CEST (Supporting Information Figures S1c and S1d). MRI were shown for the contralateral normal and ischemic lesion from a representative acute stroke rat. Fig. 1e shows the cerebral blood flow (CBF) map, revealing ipsilateral hypoperfusion.

Fig. 1.

Multi-parametric images of a representative acute stroke rat. a) Parametric T₁ map. b) Parametric T₂ map. c) Parametric ADC map. d) pH-weighted MTR_asym map obtained from CW-APT MRI. e) CBF map.

Table 1.

Comparison of multi-parametric MRI from the ipsilateral ischemic lesion and the contralateral normal area, and their difference from all stroke rats.

	Contralateral Normal Area	Ipsilateral Ischemic Area	Δ (Ipsi-Contra)	CNR
T1 (s)	1.54 ± 0.08	1.68 ± 0.05 *	0.14 ± 0.07	0.82 ± 0.36
T2 (ms)	54.81 ± 1.84	54.50 ± 1.53	−0.31± 1.32	0.37 ± 0.20
CBF (ml/g·min)	2.33 ± 0.46	0.88 ± 0.41 **	−1.45 ± 0.44	3.95 ± 1.09
ADC (μm2/ms)	0.86 ± 0.02	0.63 ± 0.05 **	−0.23 ± 0.05	4.53 ± 0.78
MMTR (CW-CEST)	0.30 ± 0.01	0.29 ± 0.01	−0.01 ± 0.01	0.33 ± 0.31
MTRasym (%) (CW-CEST)	−3.63 ± 0.63	−5.78 ± 0.38 **	−2.16 ± 0.41	2.07 ± 0.31
MRAPTR (%/s) (CW-CEST)	−0.02 ± 0.05	−1.70 ± 0.24 **	−1.68 ± 0.26	3.59 ± 0.40
MMTR (pulsed-CEST)	0.26 ± 0.01	0.25 ± 0.02	−0.01 ± 0.01	0.68 ± 0.48
MTRasym (%) (pulsed-CEST)	−3.35 ± 1.12	−5.10 ± 1.24 **	−1.74 ± 0.74	1.65 ± 0.77
MRAPTR (%/s) (pulsed-CEST)	0.03 ± 0.06	−1.35 ± 0.26 **	−1.38 ± 0.27	3.25 ± 0.65

Two sample t-test with equal variance was performed to evaluate the ischemia-specific changes.

^*P<0.05 and

^**P<0.01

The MRAPT analysis of pulse-APT MRI was shown in Fig. 2. Briefly, Fig. 2a shows mean MTR (MMTR) image with little signal change in the ipsilateral ischemic region (0.26 ± 0.01 vs. 0.25 ± 0.02, P>0.05), consistent with the observation that MTR is not pH sensitive (39). The R_1w map (Fig. 2b) shows a small signal drop in the ipsilateral ischemic lesion (from 0.65 ± 0.03 to 0.60 ± 0.02, P<0.05). The B₀ field inhomogeneity (Fig. 2c) was determined from the CW-CEST Z spectrum (27,40), displaying minor inhomogeneity. Fig. 2d shows pH-weighted MTR_asym map obtained from the pulse-APT MRI, with a signal reduction in the ipsilateral ischemic lesion (−3.35 ± 1.12% vs. −5.10 ± 1.24%, P<0.01), similar to that from CW-APT MRI (Fig. 1d). The paired t-test showed no significant MTR_asym difference from the contralateral normal tissue or the ipsilateral ischemic lesion, using CW- and pulse-APT MRI (P>0.05). The heterogeneous MTR_asym from the intact tissue can be described by MRAPT analysis using a multiple linear regression against R_1w, MMTR and B₀ inhomogeneity, per pixel (R²=0.77, Fig. 2e), following a recently proposed fast B₀ inhomogeneity correction algorithm (28). After the correction of the non-pH dependent intrinsic MTR_asym shift, the pH-specific ΔMRAPTR image (Fig. 2f), although with some small residual heterogeneity in the intact tissue, shows improved conspicuity of the ischemic lesion over the routine MTR_asym map (Fig. 2d). The CNR from pulse-APT MRI between the contralateral normal area and the ischemic lesion was 1.65 ± 0.77 and 3.25 ± 0.65 (P<0.01) for MTR_asym and ΔMRAPTR, respectively. In comparison, the CNR from CW-APT MRI was 2.07 ± 0.31 (MTR_asym) and 3.59 ± 0.40 (ΔMRAPTR, P<0.01). It is necessary to point out that the CNR difference of MTR_asym and ΔMRAPTR was not different between CW- and pulse-APT MRI (P>0.05).

Fig. 2.

pH-sensitive MRI. a) MMTR (±3.5 ppm) map from pulse-APT MRI. b) R_1w map. c) The field inhomogeneity map determined from the WASSR approach. d) pH-weighted MTR_asym map from pulse-APT MRI. e) Multivariate regression between R_1w*MTR_asym against R_1w and MMTR, per pixel from the contralateral intact tissue. f) pH-specific ΔMRAPTR map determined from the pulse-APT MRI.

Fig. 3 evaluates the dependence of pH-weighted MTR_asym and pH-specific ΔMRAPTR effect as a function of RF irradiation pulse duration (τ) and flip angle (FA). Fig. 3a shows MTR_asym difference between the contralateral normal and ischemic lesion as functions of saturation RF pulse duration (τ) and flip angle (FA). Fig. 3b compares the CNR between the normal area and ischemic lesion calculated from the MTR_asym image. The CNR peaks around FA of 180° and a pulse duration between 10 and 15 ms. The magnitude of ΔMRAPTR change between the contralateral normal and ischemic lesion and its corresponding CNR were shown in Figs. 3c and d, respectively. Similarly, the ischemia related ΔMRAPTR CNR peaks for a flip angle of 180° and duration between 10 and 15 ms. This is consistent with reports that for the slow chemical exchange rate, the optimal pulse CEST RF irradiation has a flip angle of 180° (19,20).

Fig. 3.

2D view of wireframe mesh plots for optimization of pH-weighted pulse-APT MRI. a) The magnitude of MTR_asym difference between the contralateral normal and ischemic areas as functions of RF pulse duration (τ) and flip angle. b) The MTR_asym CNR between the contralateral normal and ischemic areas as functions of τ and FA. c) The magnitude of ΔMRAPTR difference between the contralateral normal and ischemic areas as functions of τ and FA. b) The ΔMRAPTR CNR between the contralateral normal and ischemic areas as functions of τ and FA.

We compared CW- and pulse-APT MRI scans in acute stroke rats. Specifically, Figs. 4a and 4b show that MTR_asym (R²=0.85, P<0.01) and ΔMRAPTR (R²=0.82, P<0.01) derived from CW- and pulse-APT MRI were highly correlated. pH was calculated from CW-APT MRI using $\mathrm{p}\mathrm{H}=7.05+\mathrm{l}\mathrm{o}\mathrm{g}10\left(1+\frac{∆\mathrm{M}\mathrm{R}\mathrm{A}\mathrm{P}\mathrm{T}\mathrm{R}}{{\mathrm{C}}_0}\right)/{\mathrm{C}}_1$, with C₀ and C₁ being 5.04 and 0.25, respectively (25). Building on the highly correlated CW- and pulse-APT MRI results, we also estimated tissue pH from pulse-APT MRI. pH maps determined from ΔMRAPTR images from CW- (Fig. 4a) and pulse-APT MRI (Fig. 4b) agree well. The pH values from CW- and pulse-APT MRI were highly correlated (Fig. 4c). Specifically, pH of the contralateral normal tissue was 7.05 ± 0.02 and 7.07 ± 0.02 from CW- and pulse-APT MRI, respectively, with the ischemic lesion pH being 6.33 ± 0.11 and 6.28 ± 0.12, respectively. Although the paired t-test showed a significant difference (P<0.05), the magnitude of pH difference between CW- and pulse-APT MRI was very small, being −0.03 ± 0.02 for the normal tissue and −0.06 ± 0.03 for the diffusion lesion, respectively.

Fig. 4,

Comparison of pH map determined from CW- and pulse-APT MRI. a) pH map determined from MRAPTR analysis of CW-APT MRI. b) pH map determined from MRAPTR analysis of pulse-APT MRI. c) The correlation between pH determined from CW- and pulse-APT MRI, per pixel.

Discussion

Our study generalized pH-specific magnetization transfer and relaxation normalization analysis, originally developed for CW-APT imaging of acute stroke, to pulse-APT MRI. We showed that the contrast to noise ratio between the contralateral normal area and ischemic lesion increased from 1.65 ± 0.77 (MTR_asym) to 3.25 ± 0.65 (ΔMRAPTR) using pulse-APT MRI, demonstrating the generalizability of MRAPT analysis. Notably, the CNR of MRAPTR maps determined from CW- and pulse-APT MRI were not significantly different from each other (3.59 ± 0.40 vs. 3.25 ± 0.65, P>0.05). As such, highly pH-specific pulse-APT MRI is feasible with the generalized MRAPT analysis, which is promising for characterizing the heterogeneous ischemic tissue injury typically characterized by perfusion and diffusion MRI (41–44).

Although the development of omega plot-based quantitative CEST (qCEST) MRI is capable of determining the labile proton ratio and exchange rate concurrently, further work is needed to quantify multi-pool CEST effects in vivo (45–47). Because APT MRI takes a relatively long scan time, our study did not execute the full 2-dimensional optimization experiments but instead focused on two one-dimension experiments (i.e., we evaluated the pulse duration dependence for a fixed flip angle and varied the flip angle for an intermediate pulse duration.) Although the optimization experiment was not exhaustive, it has been shown that an inversion pulse train is optimal for capturing the CEST effect of slow exchanging species, consistent with prior studies (19,20). This is consistent with our finding that the optimized pulse-APT MRI is of similar CNR as that determined from CW-APT MRI during acute stroke, both MTR_asym and ΔMRAPTR measurements. It is worth noting that ΔMRAPTR (Supporting Information Figure S2b) not only has a narrower range than MTR_asym (Supporting Information Figure S2a) but also appears to condense into two clusters. This clustering pattern is due to the improved pH specificity of MRAPT analysis over MTR_asym (24). Because the intact tissue has a reasonably uniform intracellular pH, the MRAPT analysis effectively standardizes the APT contrast in vivo. Images from other acute stroke rats were included in the Supporting Information Figures S3–6.

Conclusion

Our study demonstrated pH-specific magnetization transfer and relaxation normalization of pulse-APT imaging of acute stroke. pH-specific ΔMRAPTR determined from pulse-APT MRI highly correlated with that of CW-APT MRI. Tissue pH determined from pulse-APT was in good agreement with that derived from CW-APT MRI, within 0.1.

Supplementary Material

FigS1-S6

Figure S1. Comparison of Z-spectra obtained from CW- and pulse-CEST MRI. a) CW-CEST Z-spectra from the contralateral intact (red) and ipsilateral ischemic lesion (blue) areas, respectively. b) CW-CEST MRI asymmetry spectra from the contralateral intact and ipsilateral ischemic lesion areas. c) pulse-CEST MRI Z-spectra from contralateral intact and ipsilateral ischemic lesion areas. d) pulse-CEST MRI asymmetry spectra from contralateral intact and ipsilateral ischemic lesion areas.

Figure S2. Evaluation of pulse- against CW-APT MRI. a) The correlation of MTR_asym determined from CW- and pulse-APT MRI, per pixel. b) The correlation of ΔMRAPTR determined from CW- and pulse-APT MRI, per pixel.

Figure S3. MTR_asym, ΔMRAPTR, and pH images derived from CW- and pulse-APT MRI of a representative acute stroke rat.

Figure S4. MTR_asym, ΔMRAPTR, and pH images derived from CW- and pulse-APT MRI of a representative acute stroke rat.

Figure S5. MTR_asym, ΔMRAPTR, and pH images derived from CW- and pulse-APT MRI of a representative acute stroke rat.

Figure S6. MTR_asym, ΔMRAPTR, and pH images derived from CW- and pulse-APT MRI of a representative acute stroke rat.