Evaluation of Sensitivity and Reliability of FunctionalR Spectroscopy Using Virtual Titration

Magnetic resonance spectroscopy (MRS) can be used to provide chemical information within human body noninvasively. Combining it with the concept of functional magnetic resonance in imaging (fMRI), functional magnetic resonance spectroscopy (fMRS) has become a powerful tool to investigate neural physiology. However, it has been reported that spectral line properties can be affect by blood-oxygenation-level-dependent (BOLD) effect, and the resultant impact on metabolite quantification has not yet been thoroughly evaluated. But this impact could be critical since in fMRS studies, the tiny changes of the concentrations are of interest. In this study, we simulated tiny concentration changes of creatine, choline, and N-acetyl-L-aspartic acid of 13 spectra to evaluate the sensitivity and reliability of LCModel (MRS quantification software). In addition, we simulated linewidth alterations that could result from BOLD effect, to examine whether the quantification result will be affected or not. Our result shows that concentration changes exceeding ±0.8% could be detectable for the three metabolites, but the actual concentration change may be over-estimated when quantifying creatine and choline. Besides, linewidth alterations result in consistent concentration changes for the three metabolites under the condition of high SNR. But in the presence of noise, results of concentration changes shows large variations among spectra. We conclude that LCModel is sensitive to tiny concentration changes, but it would be necessary to develop a calibration procedure to eliminate pseudo concentration changes.