A Comparison of Osprey/LCModel and FSL-MRS Fitting Results from Clinical PTSD Data: Baseline Estimation and Fitting Methods with 7T fMRS

Abstract

Magnetic resonance spectroscopy (MRS) is a powerful tool for investigating the underlying processes of brain disorders. Despite its power, there are concerns about the accuracy and reproducibility of results. These results often arise from differences in MRS fitting packages and a natural inclination of researchers only interested in brain dynamics to be indifferent to the default fitting settings. The goal of this study was to investigate how changing one of these settings can lead to different results, and how the difference in results varies across fitting software. We used previously acquired 7T MRS and functional MRS (fMRS) acquired from the dorsolateral prefrontal cortex of participants with varying levels of trauma exposure during a working memory task. Data were analyzed in Osprey/LCModel and FSL-MRS for comparison of spline flexibility. A validation of 2-dimensional (dynamic) fitting was performed separately in FSL-MRS. Overall, we found increasing glutamate concentration as spline stiffness increased in both software. Furthermore, we found more consistent results across spline flexibilities in Osprey/LCModel than FSL-MRS. Lastly, we observed that dynamic fitting increased the precision of fMRS results, but the accuracy of these results was difficult to discern. These results substantiate a need for further investigation into the differences between MRS fitting software.