SMART MRS: A Simulated MEGA-PRESS ARTifacts toolbox for GABA-edited MRS

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Computational Neuroscience

Background:

• Magnetic Resonance Spectroscopy (MRS) is crucial for in vivo metabolite quantification.
• Edited MRS data, particularly for gamma-aminobutyric acid (GABA), often suffers from artifacts that complicate analysis.
• Simulated data is valuable for developing and validating MRS data processing techniques.

Purpose of the Study:

• To develop a flexible Python-based toolbox (SMART_MRS) for simulating common artifacts in single-voxel GABA-edited MRS data.
• To provide researchers with a tool to generate diverse and realistic artifact-corrupted MRS datasets.

Main Methods:

• The SMART_MRS toolbox includes functions to simulate various artifacts: spurious echoes, eddy currents, nuisance peaks, line broadening, baseline contamination, frequency drifts, and frequency/phase shifts.
• Applied functions allow for simulating complex effects like lipid contamination and motion artifacts.
• Input/output functions support multiple data formats (MATLAB FID-A.mat, NIfTI-MRS) and data types (edited/non-edited, time/frequency domain).

Main Results:

• Simulated data from SMART_MRS were used to train a deep learning model for frequency and phase correction, demonstrating its utility on in vivo data.
• Visual assessment confirmed that the simulated artifacts closely resemble those observed in real-world in vivo MRS data.
• The toolbox effectively enhances the diversity and quality of simulated edited-MRS datasets.

Conclusions:

• SMART_MRS is an easy-to-install Python toolbox that significantly improves the simulation of artifacts for edited MRS data.
• It complements existing MRS simulation software by offering enhanced artifact simulation capabilities.
• The toolbox is valuable for developing robust MRS data processing algorithms and improving data quality in neuroscience research.