ABSTRACT An estimated 8.5 million Americans suffer from peripheral artery disease (PAD) and it not only interferes with one’s active lifestyle; but also increases the risks for heart attacks and strokes. Most PAD patients are asymptomatic, present with atypical pain, and a minority face serious consequence of a threatened limb. Once diagnosed with PAD, a range of evidence-based therapeutic and interventional options are available, which includes supervised exercise, antithrombotic medications, and restoring circulation through catheter and surgical interventions are available. These have potentials to improve the functional status of the patient and reduce the mortality. Oxidative phosphorylation (OXPHOS) capacity in skeletal muscle are known to get compromised in PAD and hence, tools (imaging or otherwise) are needed for aiding in the early identification of patients with PAD and to monitor the effect of supervised exercise and revascularization. Phosphorus (31P) magnetic resonance spectroscopy (31PMRS) has traditionally been used as a noninvasive tool to evaluate the OXPHOS capacity and the change in intracellular pH of exercised skeletal muscle. However, it suffers from low coverage, poor spatial resolution and a very high coefficient of variation (COV) (~20-30%) with respect to half-life of phosphocreatine recovery (τPCr). Our team previously developed a novel 2D MRI method, named creatine chemical exchange saturation transfer (CrCEST) imaging, as an alternative with improved spatial resolution (~1x1 mm2) and a temporal resolution (τRes) of ~30s. Recently, we further improved on this method to enable the 3D coverage while maintaining the same τRes of 30s. The estimation of creatine recovery half-time (τCr) on a voxel-wise basis is currently not feasible due to the coarse temporal resolution (30s) combined with the low effective signal to noise ratio (eSNR) (<20) of 3D-CrCEST time-series. Rather, we performed muscle-group-specific fit, where the useful information in form of a voxel-wise inherent anatomic/physiological variation were sacrificed in return for increased SNR. In another important development, we developed a spatial regularized reconstruction approach and showed the feasibility of a noise robust reconstruction, provided τCr ≳4τRes. Given that the typical τCr of healthy control is ~45-80s for our plantar flexion exercise protocol, there is a need to improve on τRes ~10-12s. In this proposal, we aim to improve on the temporal resolution: τRes of CrCEST by implementing keyhole imaging in combination with the alteration in frequency offset (FO) schemes. By aiming to improve τRes by a factor of ~2-3x, the information content present in CrCEST time series would be sufficiently increased to enable a noise robust reconstruction of τCr-map. To reduce the component of variability with physiological origins, we plan to adjust the exercise load to avoid unpredictable and drastic changes in perfusion profile. For the validation, we will compar...