BART: Tutorials, Workshops & Webinars
Figure: Simulated MRI images.
Table of contents
Getting Started with BART1. Linux/unix terminal
2. Python
3. Matlab
Non-Cartesian Reconstruction
1. Non-Cartesian SENSE reconstruction
2. GRASP reconstruction
Model-based Reconstruction
1. Subspace-constrained reconstruction
2. Model based reconstruction
Dynamic MRI Reconstruction
1. Introduction to dynamic MRI reconstruction with BART
2. Dynamic Contrast Enhanced (DCE) MRI
Deep Learning with BART
1. Introduction to machine learning reconstruction with BART
2. Machine learning reconstruction with BART
3. TensorFlow and deep learning reconstruction with BART
SENSE reproducibility challenge
The BART solution of the SENSE reproducibility challenge
More useful links
The 2020-2022 webinar series
Getting Started with BART
1. Terminal
1.1 Using BART through the linux/unix command line: webinar for new users (2020)
This two-days webinar (from June 2020) introduced the concepts of working with BART through the linux/unix command line.
Day 1
- Where to find docs, examples, and help
- Discussion of file format and dimensions
- Working with Command Line Interface (CLI) tools and Matlab/Python wrappers
- Data preprocessing
- Compressed Sensing and non-Cartesian MRI reconstruction
- GRASP-like MRI reconstruction
Day 2
- Intro to the C-programming library
- Set up a build environment, compile, and run
- Add command-line parameter to existing tool
- Build a basic tool and system test
1.2 BART workshop - ISMRM 2019
The workshop tutorial included an introduction to the BART command-line tools. Topics covered:- The basic structure of BART commands
- displaying images
- using the BART phantom tool
- viewing data dimensions
- Bitmasks
- Low-pass filtering
- Reconstruction of multi-coil data
- Reconstruction with parallel imaging compressed sensing (PICS)
1.3 BART workshop - ISMRM 2016
The workshop included demos for:- Creating multi-channel k-space data
- Using BART's Shepp Logan phantom
- Coil compression and image reconstruction
- Generating ESPIRIT sensitivity maps using BART's ecalib tool
- Using different regularization techniques
2. Python
2.1 Webinar for new users - python
The webinar we had on July 2021 includes a demo and hands-on exercise for working with BART in a python environment. It covers:
- How to create phantom data and subsampling masks in BART.
- How to use BART for parallel MRI reconstruction.
2.2 Getting started with BART - 45min tutorial
The ESMRMB MRI-Together 2021 workshop included a demo and hands-on exercise for working with BART in a python environment. It covers:
- How to set up an environment for BART.
- The BART command structure and data format.
- How to use Bitmasks to select dimensions.
- How to create a Cartesian k-space phantom.
- Example: Subspace T1 Mapping - including coil compression, trajectory generation and coil sensitivity estimation.
- Introduction to using BART for machine learning, including reconstruction with MoDL and variational networks.
2.3 Data processing and computing g-factor for parallel imaging
The ISMRM 2019 workshop included a tutorial that shows how to use the BART-Python interface for:- Loading data from h5 files
- Noise whitening
- Visualizing the sampling pattern
- Reconstruction
- Computing g-factor for parallel imaging.
3. Matlab
Matlab Interface and Examples
The toolbox can also be used in combination with Matlab/Octave.
>> sensitivities = bart('ecalib', kspace);
>> image_out = bart('pics -l1 -r0.001', kspace, sensitivities);
More examples where the tools are called directly from Matlab can be found here.
Matlab code and data: GitHub repository
A Matlab-based image viewer which works well with BART is arrayShow by Tilman Sumpf.
3.1 Inroduction to the BART-Matlab interface
The first webinar of 2020 included a demo for using the BART Matlab API.3.2 Using the BART-Matlab interface
The ISMRM 2016 workshop included a demo for performing a Wave-CAIPI reconstruction followed by a retrospective Wave-CS reconstruction tool with the BART Matlab API.Non-Cartesian Reconstruction with BART
1. Linux/unix terminal
1. Non-Cartesian SENSE reconstruction
The ISMRM 2019 workshop included a demo for non-Cartesian SENSE reconstruction2. GRASP reconstruction
The ISMRM 2016 workshop included a demo for GRASP reconstruction from golden-ratio radial k-space sampling.Model-based Reconstruction with BART
1. Subspace constraint reconstruction & model-based reconstruction
The webinar of March 2021 included demos and hands-on tutorials for:
- Subspace-constrained reconstructions
- Analytical model-based phantom simulation
2. Model based reconstruction
The ISMRM 2021 BART software tutorial included an interactive demo on nonlinear model-based reconstruction for quantitative MRI (T1 mapping, water-fat separation)
Dynamic MRI Reconstruction with BART
1. Introduction to dynamic MRI with BART
The webinar of Dec 2020 included demos and hands-on tutorials for:
- Dynamic (temporal sequence) MRI data loading, dimensions organization, and sensitivity maps computation.
- Image reconstruction using parallel-imaging-compressed-sensing (PICS), with advanced regularization methods suitable for dynamic MRI data.
2. Dynamic Contrast Enhanced (DCE) MRI reconstruction
The ISMRM 2016 software demo included a tutorial on dynamic axial-slice reconstruction with BART.
3. GRASP reconstruction
The ISMRM 2016 workshop included a demo for GRASP reconstruction from golden-ratio radial k-space sampling.Deep Learning with BART
1. Introduction to machine learning reconstruction with BART
BART webinar #6 (March 1, 2022) included demos and hands-on tutorials for working with BART in a python environment. It covers:
- Introduction to using BART for machine learning.
- TensorFlow-regularization + BART Reconstruction.
- Examples for data pre-processing.
- Examples for training MoDL and a Variational Network with BART.
2. Machine learning reconstruction with BART
The ESMRMB MRI-Together 2021 workshop included a demo and hands-on exercise for working with BART in a python environment. It covers:
- Introduction to using BART for machine learning.
- Reconstruction with MoDL and Variational Networks.
3. Tensorflow and neural networks with BART
The ISMRM 2021 BART tutorial included demos and hands-on tutorials for:
- TensorFlow-Regularizer + BART Reconstruction
- Neural networks with BART
- How to run BART on Google Collab
SENSE Reproducibility Challenge
The BART solution of the SENSE reproducibility challenge
The challenge was organized by the ISMRM. The BART solution was presented in our Dec 2020 webinar:More useful links
Our 2020-2022 webinars series can also be found here:
Webinars & workshops material in chronological order
There are new tutorials from our Webinar which you can find in a GitHub repository.
ISMRM 2021 As part of the ISMRM 2021 meeting, we gave a demo of the new features of BART related to non-linear model-based reconstruction and deep learning integration.
ISMRM 2016 The toolbox was presented at the ISMRM 2016 Data Sampling and Image Reconstruction Workshop. This material was created for BART version 0.3.00 and later versions might have minor differences. Please check the README included with each release for up-to-date installation instructions.
Demo code and data: GitHub repository