ATP is the essential energy source of myocardial contraction. The synthesis of myocardial ATP involves the conversion of phosphocreatine to creatine catalyzed by creatine kinase. It has been previously shown that cardiac dysfunction is associated with myocardial ATP loss^1,2.

Chemical exchange saturation transfer (CEST) is an emerging technique that enables detection of endogenous or exogenous metabolites. It has been used to map creatine distribution in the myocardium to assess metabolic activity in animal models³. However, the previous approach requires lengthy scan time (50 min), which needs to be reduced considerably for human application.

In this work, we developed an optimized cardiac CEST technique with dramatically shortened scan time (by 10-fold), improved motion registration and CEST signal calculation, and tested its feasibility to detect chronic myocardial infarction in porcine model and also in a patient for the first time. LGE imaging was used as reference.

Fig. 1 shows the pulse sequence diagram of the proposed cardiac CEST technique. ECG triggering was used to reduce cardiac motion artifacts. The readout was placed during the quiescent period of the cardiac cycle. Since each image was acquired by single-shot FLASH (~200 ms readout period), there were minimal respiratory motion artifacts in each image. All images at different saturation frequency offsets were acquired at the same respiratory phase via navigator gating. TR was set to be 4000 ms so that magnetization is largely recovered during each TR after data acquisition. 33 images were collected at different saturation frequency offsets ranging from -4.8 ppm to 4.8 ppm with a step size of 0.3 ppm. CEST preparation module consists of five Gaussian pulses of flip angle 2700° and duration of 30 ms at duty cycle of 50% (B_1rms is 3.76 uT). There is a spoiler gradient after each Gaussian pulse to crush the residual transverse magnetization. Spatial resolution was maintained at 2.3 x 2.3 x 8.0 mm³. CEST contrast map, a representation of creatine distribution map, was generated using pixel-by-pixel Z-spectrum fitting.

Cardiac CEST imaging technique was optimized in the following aspects: (a) Images were acquired by single-shot FLASH instead of segmented acquisition, resulting in an imaging time of 4-5 min, depending on the navigator acceptance rate. (b) All images were registered using advanced normalization tools (ANTs)⁴ to further reduce the residual respiratory motion (up to 4 mm myocardium displacement). This helps improve the robustness of the cardiac CEST technique, especially since pixel-by-pixel mapping was performed afterwards. (c) Z-spectrum was fitted to the Lorentzian-shaped 3-pool-model to generate CEST contrast map. This fitting method helps increase the reliability of the generated CEST contrast map by reducing the impact of signal fluctuation from B₀ field shifting and residual motion.

Four female Yucatan porcine and one patient with chronic myocardial infarction were studied on a 3T Siemens Verio clinical scanner. LGE images were acquired as reference for myocardial infarction.

Fig. 2(a-b) shows representative CEST contrast maps and corresponding LGE images in porcine and the patient. The hypointense region in the CEST contrast map matches the bright area in LGE image closely, suggesting that the scar region has reduced creatine distribution and lower metabolic activity compared to healthy myocardium. This is consistent with previous study⁵.

Fig. 2(c) quantitatively compares the CEST signals in the LGE positive and negative regions in base, mid and apex slices in the porcine model. The CEST signal is significantly reduced in the infarct region (9.5%±1.9%), compared to healthy remote myocardium (15.5%±2.2%), p<0.00005. In the patient, CEST signal in the infarct region is 8.4% while that in the healthy myocardium is 16.2%.