Compressed sensing data acquisition and reconstruction for fast phosphorus magnetic resonance spectroscopic imaging

Brain is the most important organ for the human life, and therefore, brain tumors are among the tumors that are vital to diagnose, follow-up and treat. Magnetic Resonance Imaging (MRI) is the most commonly used imaging modality for diagnosis of brain tumors. Phosphorus MR spectroscopic imaging (31P-MRSI) is a non-invasive metabolic imaging modality that can provide quantitative information about metabolism, energy and oxygen state, membrane synthesis and degradation, and pH of the tissue. 31P-MR spectroscopic imaging cannot be practically used in clinical applications due to its extended acquisition times to enhance low phosphorus MR signal, despite all the important information it provides. To accelerate the acquisition process of MR, compressed sensing method has recently been used. Due to its denoising effect, compressed sensing does not reduce the signal to noise ratio (SNR). This study aims to use compressed sensing to reduce the acquisition time of 31P-MRSI in clinical applications. Initially, required parameters and undersampling patterns were analyzed and optimal values were determined for accelerating the 31P-MRSI data acquisition with compressed sensing, using simulations. Using Philips Paradise environment, a 31P-MR spectroscopic imaging pulse sequence was implemented for accelerating the data acquisition with compressed sensing, and this pulse sequence was integrated into a 3T Philips MR scanner. A volunteer and a brain tumor patient, who gave their informed consent, were scanned. The signal to noise ratio, peak heights and ratios were compared between accelerated and full k-space data using Bland Altman statistical test. Using Wilcoxon rank sum test, differences of spectral peak heights and ratio between the healthy and tumor tissue were observed for both full and compressed sensing datasets. This study showed that compressed sensing reconstruction could be used to accelerate 31P-MR spectroscopic imaging of human brain and brain tumors.