The repository provides the source code for "A Deep Learning Approach for Fast Muscle Water T2 Mapping with Subject Specific Fat T2 Calibration from Multi-Spin-Echo Acquisitions"

Full citation: Barbieri, M., Hooijmans, M.T., Moulin, K. et al. A deep learning approach for fast muscle water T2 mapping with subject specific fat T2 calibration from multi-spin-echo acquisitions. Sci Rep 14, 8253 (2024). https://doi.org/10.1038/s41598-024-58812-2

The deep learning application has been developed in TensorFlow. TensorFlow, Numpy, and Scipy are required to run the deep learning application.

The EPG code for simulating the Multi-Echo-Spin-Echo (MESE) data is provided in Matlab. The core EPG function is taken from the StimFit toolbox (https://github.com/usc-mrel/StimFit). Since the actual pulse shapes used by the vendors in implementing the MESE sequence are proprietary information, the simulations are provided for SINC pulses with TBW = 2.

The functions used to generate the excitation and refocusing pulses and respective slice profiles are provided and make use of the RF Pulse Design toolbox created by John Pauly (https://rsl.stanford.edu/research/software.html)

The raw MESE data (in NIfTI format) and processed data can be downloaded from Zenodo (https://doi.org/10.5281/zenodo.10520542)

Conventional EPG fitting algorithms based on Dictionary and Non-Linear Lease-Squared (NLSQ) methods are also provided in Matlab.