³¹P MRS has been used to quantify energy metabolism by means of dynamics of tissue PCr and pH kinetics for at least 30 years. A high spatial and temporal resolution is desired to capture the dynamics of sever tissues¹. CSI suffers from slow acquisition speed while single voxel MRS is limited in tissue coverage. Therefore, ³¹P MRI has been introduced to quantify PCr kinetics in exercise recovery², but so far, pH values derived from MRI have not been reported. We propose a multiply spectrally selective gradient-echo MRI pulse sequence that is capable of measuring both PCr and P_i with their respective chemical shifts.

Time-interleaved acquisition of spectrally selective k-space lines was inserted in a low flip-angle 3-D gradient-echo pulse sequence (Figure 1). The phase difference Φ between the set of images is T_E times the chemical shift Ω (figure 2), Therefore, by acquiring N sets of images at different resonance frequencies, N metabolite images can be acquired quasi simultaneously (shifted by T_R/N). pH can be calculated from the chemical shift between PCr and P_i.

10 healthy volunteers (2m/8f) were studied using a Siemens 7 T scanner and a Rest - Exercise – Recovery protocol. Within the study, adaptations were made to increase the mechanical duty cycle (exercise duration between 3 and 5 min)³ and speed up measurements:

Subject 1-6: 16x16x6 2 ml voxels, acquisition time 5.8 s, 4.2 s delay for 2 plantar flexions.

Subject 7-10: 16x16x4 3 ml voxels, acquisition time 3.8 s, 5.2 s delay for 3 plantar flexions.

Acquisition parameters: $$$T_R=60 ms$$$, 2 acquisitions at Ω=(0,4.8) ppm, T_E=3.8 ms, RF pulse: 5 ms/350 Hz. Short echo times are required for P_i (T₂^* decay), therefore a relatively high receiver bandwidth (280 Hz/px) was used.

ROI-based analysis of images corresponding to gastrocnemius medialis, lateralis and soleus muscles was performed. Magnitude images give bias in low SNR data (Figure 3), therefore real-value absorption mode images were used for quantifying PCr kinetics. PCr was integrated over the ROI and fitted mono-exponentially during recovery. The chemical shift and hence pH was calculated from temporally unwrapped phase of the ROI-integrated complex data. Analysis was performed using in-house developed software using PDL (http://pdl.perl.org) and Prima toolikt (http://www.prima.eu).

To validate the data, MRS data (single-voxel semi-LASER) was also acquired in gastrocnemius medialis using the same exercise protocol in the same scan session. Spectra were quantified using jMRUI/AMARES. PCr recovery was fitted mono-exponentially.

Results are given as mean and standard deviation. Intra-subject comparisons were done using a paired student's t-test.

Both PCr and P_i gradient-echo images with high temporal and spatial resolution were acquired. PCr was readily visible without averaging in as little as 3.8 s, P_i has a low concentration (and signal) at rest (Fig. 4b). It was only visible in voxels of muscles involved in exercise (Figure 4d). PCr depletion was higher in gastrocnemius (medialis: 43±13%, lateralis: 48±17 %) than in soleus (20±11%). The recovery time τ_PCr could be fitted when the depletion was higher than 20% (gastrocnemius 10 subjects; medialis 75.7±22.1 s, lateralis 93.1±40.8 s, soleus 6 subjects; 75.6±37.9 s). In accordance with prior experience and spectroscopy data (Table 1), pH (Figure 4e) rose initially in all muscles and subsequently declined more in stronger exercising gastrocnemius (medialis 6.74±0.18, lateralis 6.65±6 0.27) than in soleus (6.96±0.12) during exercise. When averaged over several images, even resting pH can be calculated for a given ROI (Table 1).

In this study, we present a frequency-selective fast 3-D gradient-echo sequence for simultaneous acquisition of signals from multiple metabolites. The phase images contain the spectral information. Signal dependent noise distribution in magnitude images might introduce a systematic error and thus bias the quantification of PCr towards lower depletions and longer recovery times. Therefore, real part images, with symmetric and signal-independent noise distribution, were used for quantification of PCr dynamics.

At the chosen echo-time, phase wraps do not occur over the physiological pH range. Some bias from PDE or ATP signals could introduce some error, negligible under most conditions. The point-spread function of low resolution MRI is worse than the semi-LASER voxel profile, probably explaining the discrepancy in PCr depletion.

In conclusion, we report that the combined high temporal and spatial resolution achieved with the designed ³¹P-MRI sequence represents a valuable low-SAR alternative to MRS for simultaneous PCr and P_i imaging during exercise-recovery experiments. The simultaneous and rapid measurements of both images enable the calculation of intracellular pH during the dynamic experiments, and will potentially allow for the identifying localised injuries, myopathies or functional deficits in peripheral arterial disease.