MRSCloud

MRSCloud simulates vendor-specific and sequence-specific basis sets for Magnetic Resonance Spectroscopy (MRS) to support spectral quantification, sequence evaluation, and cross-platform validation.

Key Features:

• Cloud-based spectral simulation: Performs spectral simulations and basis-set generation using cloud computing resources.
• Vendor and sequence specificity: Generates vendor-specific and sequence-specific basis sets for GE, Philips, and Siemens scanners.
• Supported sequences and field strength: Supports conventional acquisitions and spectral editing schemes such as PRESS and semi-LASER localization at 3 Tesla.
• FID-A integration: Built upon the spectral simulation functionality of the FID-A software package.
• Computational speedups: Integrates a one-dimensional projection method, coherence pathways filters, and precalculation of propagators to accelerate simulations.
• RF waveform generation: Generates RF waveforms based on generic pulse shapes and timings provided by vendors.
• Spatially resolved simulations: Performs simulations across spatial grids at 21×21, 41×41, and 101×101 resolutions.
• Comparative validation: Compares simulated metabolite basis functions with FID-A, MARSS, and a phantom-acquired LCModel basis set using intraclass correlation coefficients (ICCs).
• Reported concordance: Reports ICCs of at least 0.98 versus FID-A, at least 0.96 versus MARSS, and an ICC range of 0.68 (glutamine) to 0.96 (N-acetylaspartate) versus acquired LCModel basis spectra.
• Statistical analysis environment: Uses R in RStudio for statistical analyses.
• Simulation runtime: Simulates a full PRESS basis set in approximately 11 minutes on the server.

Scientific Applications:

• Basis-set generation for spectral quantification: Produces metabolite basis functions for use in spectral fitting (e.g., LCModel).
• Cross-platform comparison and validation: Enables comparison of simulated basis sets to FID-A, MARSS, and phantom LCModel basis sets using ICCs to assess concordance.
• Sequence and pulse evaluation: Evaluates effects of pulse shapes and timings, including PRESS and semi-LASER localization at 3T, on simulated spectra.
• Spatial-resolution assessment: Assesses spatial resolution effects by simulating grids at 21×21, 41×41, and 101×101.

Methodology:

Computations leverage FID-A spectral simulation functions augmented with a one-dimensional projection method, coherence pathways filters, and precalculation of propagators; RF waveforms are generated from vendor-provided pulse shapes and timings; comparisons use intraclass correlation coefficients and statistical analyses are performed in R (RStudio).