This repository contains materials and tools to support the implementation and use of the "Free Waveform" (FWF) MRI pulse sequence. The sequence is a diffusion-weighted spin-echo that facilitates the execution of user-defined gradient waveforms for the purposes of tensor-valued diffusion encoding and other methods that require arbitrary modulation of the gradients.
Siemens
Please contact Filip Szczepankiewicz at filip.szczepankiewicz@med.lu.se.
Note that the sequence is shared through Lund University by establishing:
- C2P between Siemens Healthcare and the receiver,
- MTA (material transfer agreement) between Lund University and the receiver.
Check the list of compiled variants to see if the sequence is available for your system. In special cases we may compile the sequence for other versions.
Philips
Please contact Maarten Versluis at Philips Healthcare (maarten.versluis@philips.com).
GE
Please contact Timo Schirmer at GE Healthcare (timo.schirmer@med.ge.com).
United Imaging
Please contact Weiguo Zhang at United Imaging (weiguo.zhang@united-imaging.com).
Bruker
An implementation for TopSpin, by Daniel Topgaard at Lund University, is available here.
An implementation for ParaVision, by Mathew Budde at Medical College of Wisconsin, is available here.
Siemens
Instructions for sequence installation and setup are found here.
Philips, GE and United Imaging
Instructions for installation and setup are provided by the vendor.
Designing the experiment [Review paper]
The design of the gradient waveforms (b-tensor shapes) and the signal sampling schemes (b-values, rotations etc.) must be considered when setting up he experiment. A comprehensive review of the factors that need be considered is found here. In general, the design is informed by the hardware, the intended analysis technique and the organ/subject characteristics. Below, we have collected tools and examples related to the experimental design.
Waveform design
A framework for numerical gradient waveform optimization was published by Sjölund et al. and is available on GitHub. This framework also includes concomitant gradient compensation, motion encoding compensation, as well as cross-term compensation.
Example sampling schemes
Examples of sampling schemes appropriate for a given combination of organ and analysis technique are found in the SamplingSchemes folder.
Postprocessing can be done using regular tools developed by the diffusion MRI community. Special care is however needed for correction of distortions due to eddy currents and subject movement to avoid artefacts (see Nilsson et al., 2015).
This can be done with e.g. the mddMRI framework and eddy tool from FSL although special conditions apply (see this note).
We have published an extensive framework in open source for the analysis of data encoded by b-tensors and more. Please refer to these instructions for the setup of analysis pipelines, and the interpretation of model parameters.
Example of analysis pipeline
- A brief example of how to calculate QTI parameters from data (based on the DIB2019 data set) can be found here.
- A thorough, step-by-step, example including motion correction, QTI parameter fitting, and co-registration with anatomical sequences can be found here
Tensor-valued diffusion encoding often requires complex gradient wavefomrs. In the design of these, features such as physiology, hardware, and diffusion physics must be considered. This paper reviews the major factors that go into the waveform design, and the tradeoffs that are considered.
F. Szczepankiewicz, C-F. Westin, M. Nilsson. Gradient waveform design for tensor-valued encoding in diffusion MRI. Journal of Neuroscience Methods 348, 2020.
Numerical Optimization of gradient waveforms (NOW) is an open source MATLAB framework for flexible generation of waveforms that enable q-space trajectory imaging (QTI) for tensor-valued diffusion encoding.
Sjölund, J., Szczepankiewicz, F., Nilsson, M., Topgaard, D., Westin, C. F., & Knutsson, H. (2015). Constrained optimization of gradient waveforms for generalized diffusion encoding. Journal of Magnetic Resonance, 261, 157-168.
Maxwell-compensated gradient waveforms and tools for concomitant gradient analysis [GitHub] [Citation] [YouTube]
Concomitant gradient effects can introduce gross bias in dMRI measurements. By limiting the Maxwell index in numerical optimization we can remove these errors. The concomitant field analysis (CFA) tool facilitates analysis of Maxwell terms in arbitrary gradient waveforms.
Szczepankiewicz F, Westin, C‐F, Nilsson M. Maxwell‐compensated design of asymmetric gradient waveforms for tensor‐valued diffusion encoding. Magn Reson Med. 2019;00:1–14. https://doi.org/10.1002/mrm.27828
Bulk motion and incoherent ballistic motion can be misinterpreted as diffusion. This framework extends numerical optimization of waveforms to include nulling of motion encoding to arbitrary order.
F Szczepankiewicz, J Sjölund, E Dall’Armellina, S Plein, J E Schneider, I Teh, and C-F Westin. Motion-compensated gradient waveforms for tensor-valued diffusion encoding by constrained numerical optimization. Magn Reson MEd, 2020
Imperfect magnetic fields contain background gradient that corrupt the desired diffusion encoding. This framework yields waveforms that are compensated for cross-terms with the background and we propose a method for estimating the background explicitly and remove its effects.
Szczepankiewicz and Sjölund, Cross-term-compensated gradient waveform design for tensor-valued diffusion MRI. Journal of Magnetic Resonance, 2021.
Multidimensional diffusion MRI (MD-dMRI) framework is an open source MATLAB repository that facilitates analysis of data acquired with tensor-valued diffusion encoding and its correlation with relaxation weighting.
M. Nilsson, F. Szczepankiewicz, B. Lampinen, A. Ahlgren, J. de Almeida Martins, S. Lasic, C-F. Westin, D. Topgaard. An open-source framework for analysis of multidimensional diffusion MRI data implemented in MATLAB. Proc. Intl. Soc. Mag. Reson. Med. 26 (2018), Paris, France.
This open source framework contains a MATLAB implementation of the SAFE model by Hebrank and Gebhardt which can be used predict PNS in Siemens MRI systems based on any given gradient waveform and hardware configuration.
Szczepankiewicz F, Westin, C-F, Nilsson M. Maxwell-compensated design of asymmetric gradient waveforms for tensor-valued diffusion encoding. Magn Reson Med. 2019;00:1–14. https://doi.org/10.1002/mrm.27828
The free waveform header tools are used to encode and decode information that is specific to the FWF sequence. The code provides an abstraction to ENCODE blocks of typed vectors in base64. This can be used to store floating point waveforms (and any other data type) in an efficient manner. The code also includes corresponding decoders with prototype implementation in PYTHON and MATLAB. A specific implementation exists for the FWF sequence (v1.12 and later) developed by FSz for Siemens MRI systems.
F Szczepankiewicz, S Hoge, C-F Westin. Linear, planar and spherical tensor-valued diffusion MRI data by free waveform encoding in healthy brain, water, oil and liquid crystals. Data in Brief (2019), DOI: https://doi.org/10.1016/j.dib.2019.104208
The repository contains detailed information about experimental setup, waveforms, sampling schemes and fit setting for DIVIDE or QTI at 1.5T, 3T and 7T scanners as part of the publication by Szczepankiewicz et al. 2019 in PoNE.
Szczepankiewicz F, Sjölund J, Ståhlberg F, Lätt J, Nilsson M. Tensor-valued diffusion encoding for diffusional variance decomposition (DIVIDE): Technical feasibility in clinical MRI systems. PLoS ONE. 2019;14(3):e0214238. https://doi.org/10.1371/journal.pone.0214238
This is an open source repository that supplies diffusion-MRI data with tensor-valued diffusion encoding. Data is available in a healthy human brain in vivo as well as water, oil and liquid crystal phantoms. The repository also contains detailed information and resources concerning the experiment and its design.
F. Szczepankiewicz, S. Hoge, C-F. Westin. Linear, planar and spherical tensor-valued diffusion MRI data by free waveform encoding in healthy brain, water, oil and liquid crystals. Data in Brief (2019), DOI: https://doi.org/10.1016/j.dib.2019.104208
The talk "'Fat' B-tensors and Diffusion Tensor Distributions" was presented at a conference at Cardiff University entitled "A spin thro’ the history of restricted diffusion MR" on January 31st and February 1st 2017. The conference was hosted by the Cardiff University Brain Research Imaging Centre and was sponsored by Siemens Healthineers and the EPSRC.
F. Szczepankiewicz, D. van Westen, E. Englund, C-F. Westin, F. Ståhlberg, J. Lätt, P.C. Sundgren, M. Nilsson. The link between diffusion MRI and tumor heterogeneity: Mapping cell eccentricity and density by diffusional variance decomposition (DIVIDE). NeuroImage 142, p. 522-532, 2016. DOI: https://doi.org/10.1016/j.neuroimage.2016.07.038