Logo der Physikalisch-Technischen Bundesanstalt

Open Source

Einleitung

Der Fachbereich beschäftigt sich mit der Entwicklung von Methoden im Bereich der Magnetresonanztomographie. Damit diese Methoden auch eine breite Anwendung finden und als Referenzverfahren zur Verfügung stehen, sind wir bestrebt, möglichst viele unsere Entwicklungen als open-source Software bereitzustellen. Zusätzlich stellen wir auch aufgenommen MR-Daten zur Verfügung. Die jeweiligen Lizenzbestimmungen finden sich auf den unten angeführten Links. Falls Sie die bereitgestellten Daten oder Programme hilfreich finden, würden wir uns über eine Zitierung der dazugehörigen Publikationen freuen.

8.12 In vivo MRI

Open Code

Repeatability and Reproducibility Calculation by Restricted Maximum Likelihood Analysis and Bland-Altman Analysis of Spectral Shape

Code for the Restricted Maximum Likelihood (REML) Analysis to generate standard deviations for different repeatability/reproducibility scenarios for unbalanced, nested study designs. Furthermore, the code for a Bland-Altman Analysis of the spectral shape can be found, that allows to assess repeatability/reproducibility of spectral data independently from fitting algorithms. https://gitlab1.ptb.de/LRiemann/repeatability_reproducibility

Open Data

NeuroMET - SPECIAL MRS Reproducibility

MRS raw data acquired from nine healthy volunteers in four different scan blocks on two different days using 3 different pulse variations to assess the repeatability and reproducibility of MR spectroscopy using a SPECIAL localization at 7T.

Object/Format: in vivo MRS of the posterior cingulate cortex / Siemens .dat Files https://zenodo.org/record/5500320

8.13 Quantitative MRT

Open Code

XT,YT-UNet: Undersampling Artefacts Reduction with Limited Training Data

A convolutional neural network (CNN) which can be used to reduce undersampling artefacts in accelerated 2D cardiac radial cine MRI. Because the network decomposes the 3D problem into smaller 2D sub-problems, the network is particularly suitable for situations where only limited training data is available. https://github.com/koflera/XTYT-CNN

Adaptive DL-MRI: Accelerated Cardiac MR Image Reconstruction using adaptive Dictionary Learning and Sparse Coding

Rekonstruction of 2D cardiac cine MR images using adaptive Dictionary Learning and adaptive Sparse coding as regularization methods. The number of basis functions in the dictionary as well as the optimal sparsity level are estimated online during the reconstruction and hence do not have to be chosen a-priori. The reconstruction software was developed at PTB and at the University of Innsbruck. https://github.com/koflera/AdaptiveDLMRI

DynamicRadCineMRI: Accelerated Cardiac MR Image Reconstruction using an Iterative End-To-End Trainable Convolutional Neural Network

The implementation of an end-to-end trainable iterative network which can be used to reconstruct dynamic 2D radial k-space data. The network consists of a sequence of alternating blocks given by a CNN-module and a conjugate gradient-module which incorporates the forward and the adjoint models and thus incorporates the physical knowledge about the problem. https://github.com/koflera/DynamicRadCineMRI

BMCTool: Bloch-McConnell simulation toolbox

A python package (PyPi: BMCTool) to simulate the evolution of magnetization in various chemically exchanging environments under arbitrary radio-frequency (RF) irradiation. The tool solves the Bloch-McConnell (BMC) equations using a matrix exponential approach and was originally developed to simulate Chemical Exchange Saturation Transfer (CEST) data. https://github.com/schuenke/BMCTool

Pulseq-CEST: Towards reproducibility in CEST MRI

Pulseq-CEST is a project initiated at the Max-Planck-Institute for Biological Cybernetics to facilitate reproducibility in CEST MRI research using the open-source sequence standard pulseq. The project provides a straightforward approach to standardize, share, simulate and measure different CEST preparation schemes. The python implementation has been developed at PTB. https://pulseq-cest.github.io/

CCP-SyneRBI

Open-source software project dedicated to synergistic image reconstruction. The modalities include PET, MR and CT, with future work aimed to also include SPECT. Directed by UCL with participation of PTB. http://www.ccpsynerbi.ac.uk/

Open Data

3D GRPE Resolution Phantom

MR raw data acquired of a standard resolution phantom with a 3D MR golden-angle radial phase encoding (GRPE) trajectory using a spoiled gradient echo sequence.

Object/Format: standard ACR resolution phantom / ISMRMRD https://doi.org/10.5281/zenodo.4600937

2D GRAPPA Resolution Phantom

MR raw data acquired of a standard resolution phantom with a 2D Cartesian trajectory using a spoiled gradient echo sequence.

Object/Format: standard ACR resolution phantom / ISMRMRD https://doi.org/10.5281/zenodo.2633785

Simulated motion-resolved [18F]NaF PET/MR data

This data set contains numerical simulations of a 3D simultaneous PET/MR acquisition with either respiratory or cardiac motion.

The simulated tracer is [18F]NaF with uptake in the coronary artery. Ground truth motion vector fields are also provided.

Object/Format: numerical simulation / ISMRMRD + Interfile + Nifty https://doi.org/10.7795/710.20201113

8.14 Ultrahochfeld-MRT

Open Code

Calibration-free pTx of the human heart at 7T via 3D universal pulses

MATLAB implementation to load 31 3D B1+ datasets of the human body at 7T and to compute and evaluate tailored and universal pulses in the human heart based on a library of 22 B1+ datasets and 9 unseen test cases as described in: Aigner, CS, Dietrich, S, Schaeffter, T, and Schmitter, S, Calibration-free pTx of the human heart at 7T via 3D universal pulses, submitted to Magn. Reson. Med. 2021. The channel-wise B1+ datasets of the human body at 7T are available also at the Open Data section. https://github.com/chaigner/UP_body

Respiration specific and respiration robust 3D kT point pulses for 7T cardiac imaging

MATLAB implementation to load 3D B1+ datasets of the human body acquired during deep breathing at 7T and to compute and evaluate tailored non-selective respiration specific and respiration robust kT-points pTx pulses in the human heart as described in: Aigner, CS, Dietrich, S, Schaeffter, T, and Schmitter, S, Calibration-free pTx of the human heart at 7T via 3D universal pulses, submitted to Magn. Reson. Med. 2021. The channel-wise (deep breathing) B1+ datasets of the human body at 7T are available at the Open Data section. https://github.com/chaigner/tailored-RRob

Open Data

31 shallow breathing 3D B1+ maps of the human body at 7T

31 shallow breathing 3D B1+ maps of the human body at 7T. Used in: Aigner, CS, Dietrich, S, Schaeffter, T, and Schmitter, S, Calibration-free pTx of the human heart at 7T via 3D universal pulses, submitted to Magn. Reson. Med. 2021

Object/Format: zip-file with 31 *.mat files https://doi.org/10.6084/m9.figshare.14778345.v2

10 deep breathing 3D B1+ maps of the human body at 7T

10 deep breathing 3D B1+ maps of the human body at 7T. Used in: Aigner, CS, Dietrich, S and Schmitter, S, Respiration induced B1+ changes and their impact on universal and tailored 3D kT point pulses for 7T cardiac imaging, submitted to Magn. Reson. Med. 2021

Object/Format: zip-file with 10 *.mat files https://doi.org/10.6084/m9.figshare.15172899.v1