Multi-parametric T₁ and T₂ with mSASHA

This repository provides a reference implementation of the 3-parameter mSASHA model proposed by Chow K et al in Improved accuracy and precision with three-parameter simultaneous myocardial T₁ and T₂ mapping using multiparametric SASHA. A 4-parameter model as described by Akçakaya M et al in Joint Myocardial T₁ and T₂ Mapping Using a Combination of Saturation Recovery and T₂-preparation.

Input Data

This app takes as input a series of T₁- and T₂-weighted images and outputs the corresponding T₁ and T₂ maps calculated using a 3-parameter or 4-parameter model. Images must be from a single 2D slice with preparation times stored in the MRD XML Header. Specifically, the following userParameters must be present:

• timeToCenter is a <userParameterDouble> that specifies the time (in ms) from the start of the single-shot image acquisition to the center line of k-space
• TS_x is a <userParameterDouble> that specifies the saturation recovery time for the xth image, where x is between 1 and the total number of images. For T₁-weighted images, TS is defined as the time from the end of the saturation pulse to the beginning of the single-shot acquisition. For T₁- and T₂-weighted images, TS is defined as the time from the end of the saturation pulse to the beginning of the T₂-preparation module. For non-prepared (anchor) images, TS is a very large number.
• TE_x is a <userParameterDouble> that specifies the T₂-preparation time for the xth image, where x is between 1 and the total number of images. For images without T₂ preparation, the value is 0.

Supported Configurations

This app supports 2 configs:

• mSASHA calculates T₁- and T₂ maps using a 3-parameter model.
• jointt1t2_4p calculates T₁- and T₂ maps using a 4-parameter model.

Running the app

The MRD app can be downloaded from Docker Hub at https://hub.docker.com/r/kspacekelvin/msasha-mrd-app. In a command prompt on a system with Docker installed, download the Docker image:

docker pull kspacekelvin/msasha-mrd-app

Start the Docker image and share port 9002:

docker run --rm -p 9002:9002 kspacekelvin/msasha-mrd-app

In another window, use an MRD client such as the one provided from the python-ismrmrd-server:

Run the client and send the data to the server. For 3-parameter model fitting:

python3 client.py -o mSASHA_3p_maps.h5 -c mSASHA mSASHA_img.h5

For 4-parameter model fitting:

python3 client.py -o mSASHA_4p_maps.h5 -c jointt1t2_4p mSASHA_img.h5

The output file (e.g. mSASHA_3p_maps.h5) contains the T₁ and T₂ maps (in order) stored in a single series.

Building the App

This code is intended for use with the matlab-ismrmrd-server, which implements an MRD App compatible interface using the MRD data format. The server can be run on any MATLAB-supported operating system, but Docker images can only built when running on Linux.

1. Clone (download) the matlab-ismrmrd-server repository.

   git clone https://github.com/kspaceKelvin/matlab-ismrmrd-server.git
   

2. Clone (download) this repository.

   git clone https://github.com/kspaceKelvin/mSASHA-MRD-App.git
   

3. Merge the MATLAB code from this repository into the main repository. Note: the server.m file will be overwritten.

   cp mSASHA-MRD-App/*.m matlab-ismrmrd-server/
   

4. In the MATLAB command prompt, add the matlab-ismrmrd-server folder and its sub-folders to the path.

   addpath(genpath('matlab-ismrmrd-server'))
   

5. In the MATLAB command prompt, start the server

   fire_matlab_ismrmrd_server
   

6. Send data to the server using the client (see above) to verify the code is correctly installed.

7. Compile the server as a standalone executable and build the Docker image:

   res = compiler.build.standaloneApplication('fire_matlab_ismrmrd_server.m', 'TreatInputsAsNumeric', 'on')
   opts = compiler.package.DockerOptions(res, 'ImageName', 'mSASHA-mrd-app')
   compiler.package.docker(res, 'Options', opts)