FSL-MRS

FSL-MRS performs end-to-end magnetic resonance spectroscopy (MRS) data processing, density-matrix spectral simulation, spectral fitting, and metabolite quantitation for neuroimaging and biochemical analysis.

Key Features:

• Modular Design: Structured as modular components to enable selective use and extensibility of processing and analysis steps.
• Data Compatibility: Operates on NIfTI-format data supporting single-voxel and multi-voxel spectroscopy with spatial orientation metadata.
• Pre-processing Capabilities: Implements coil-combination, frequency and phase alignment, and filtering of spectroscopic data.
• Spectral Simulation: Provides a density matrix simulation program that generates basis spectra from simple text-based descriptions of pulse sequences.
• Advanced Fitting Techniques: Performs spectral fitting as a linear combination of basis spectra with Markov chain Monte Carlo (MCMC) optimization to estimate full posterior distributions of metabolite concentrations.
• Validation and Accuracy: Validated on simulated data, phantom data, and in vivo human datasets at 3T and 7T with reported median errors (11.9% for simulations, 3.4% for phantoms) and correlations with LCModel ranging from 0.53 to 0.81.

Scientific Applications:

• Neuroimaging Metabolite Quantitation: Quantitative analysis of brain metabolites and biochemical markers in studies of neurological conditions and brain metabolism.
• Spectroscopic Acquisition Research: Support for single-voxel and multi-voxel MRS studies and integration with emerging spectroscopic acquisition techniques.

Methodology:

Data are converted to NIfTI; pre-processing includes coil-combination, frequency and phase alignment, and filtering; basis spectra are generated via density matrix simulations from text-based pulse-sequence descriptions; spectral fitting uses linear combination of basis spectra with MCMC to estimate posterior distributions of metabolite concentrations; validation was performed on simulated, phantom, and in vivo human datasets at 3T and 7T.