Skeletal Muscle of rats imaged using CEST MRI technique at preclinical 4.7T Varian scanner (Control and SOD1 G93A)

Abstract\u2014 Phosphocreatine (PCr) is a crucial substrate of muscle metabolism, and accurate quantification of its concentration is essential for diagnosing and understanding various muscular and neuromuscular diseases. Chemical exchange saturation transfer (CEST) MRI has significant potential for detecting proton exchange between PCr and water. However, achieving accurate quantification of PCr concentration (fs) and its exchange rate (ksw) with water using CEST, particularly at low fields, remains challenging due to significant overlapping confounding effects in tissues when using conventional fitting approaches. Deep learning (DL) presents a promising alternative, yet traditional DL models often struggle to capture subtle variations in the PCr CEST effect induced by changes in fs or ksw. Additionally, these models are typically trained on either fully synthetic data, which may not adequately mimic tissues, or in vivo data which lack ground truth. This study introduces a global-local two-branch DL model designed to effectively eliminate confounding effects and capture subtle variations in the PCr CEST effect. Furthermore, it employs partially synthetic data, which offers greater fidelity than fully synthetic data while retaining simulation flexibility and providing ground truth, for effective training. Experiments using digital phantoms demonstrate that this approach surpasses all other fitting methods and other combinations of DL models and training data. Moreover, when applied to in vivo healthy and amyotrophic lateral sclerosis (ALS) rat skeletal muscle, the proposed method reveals a significant reduction in PCr fs in ALS rats, which other methods fail to detect, underscoring the model\u2019s capability to identify subtle pathological changes.