Tools for aligned reconstructions of volumetric MRI data
This repository provides tools to implement the methods and reproduce the experiments included in the manuscript ''Motion corrected MRI with DISORDER: Distributed and Incoherent Sample Orders for Reconstruction Deblurring using Encoding Redundancy'', L Cordero-Grande, G Ferrazzi, RPAG Teixeira, J O'Muircheartaigh, AN Price, and JV Hajnal, arXiv:1910.00540, 2019.
The code has been developed in MATLAB and has the following structure:
contains the functions to run the illustrations and experiments included in Figs. 3-8 of the manuscript: fig[0304,0506,0708].m.
contains template functions to implement the DISORDER reorderings in MRI scanners: electrostaticRepulsionDISORDER.m, samplingDISORDER.m.
contains functions for IO, visualization, and parameter setting: disorderAlgorithm.m, extractOrthogonalPlanes.m, generateNIIFileName.m, readNII.m, visMotion.m, visReconstruction.m, visResiduals.m, visSegment.h, visTrajectory.m, writeData.m, writeNII.m, writeRaw.m.
contains external MATLAB tools.
from https://www.mathworks.com/matlabcentral/fileexchange/23629-export_fig
from https://gist.github.com/fasiha/190203eac467d8b7f9ab2c83c3b3011e.
from https://uk.mathworks.com/matlabcentral/fileexchange/8797-tools-for-nifti-and-analyze-image
from https://uk.mathworks.com/matlabcentral/fileexchange/39664-subtightplot
contains functions that implement generic methods for reconstruction: aplGPU.m, blockGPU.m, build1DFTM.m, buildFilter.m, buildFoldM.m, buildHarmonicSampling.m, buildStandardDFTM.m, cdfFilt.m, compressCoils.m, compressMotion.m, computeROI.m, convertRotation.m, downsampleOperators.m, extractROI.m, dynInd.m, fctGPU.m, fct.m, fftGPU.m, filtering.m, flipping.m, fold.m, generateGrid.m, generateTransformGrids.m, groupwiseVolumeRegistration.m, ifctGPU.m, ifct.m, ifftGPU.m, ifold.m, ind2subV.m, indDim.m, mapMat.m, margosianFilter.m, matfun.m, mirroring.m, mirrorModulation.m, multDimMax.m, multDimMea.m, multDimMed.m, multDimMin.m, multDimSum.m, normm.m, numDims.m, parUnaFun.m, plugNoise.m, precomputeFactorsSincRigidTransform.m, pyramidPlan.m, removeOverencoding.m, resampling.m, resPop.m, resSub.m, restrictTransform.m, shearing.m, shifting.m, sincRigidTransformGradient.m, sincRigidTransform.m, sub2indV.m, upsampling.m, wrapToPiHalf.m.
contains functions to perform aligned reconstructions: CGsolver.m, computeEnergy.m, constrain.m, CSsolver.m, decode.m, encodedecode.m, encode.m, LMsolver.m, precondition.m, regularize.m, solveC.m, solveG.m, solveXT.m, stopCondition.m, traceEstimation.m.
contains functions that build the shearlet frames for regularization: buildDirectionalFilter.m, buildQuadratureMirrorFilter.m, buildShearlet.m, buildWedgeBandpassAndLowpassFilters.m, checkFilterFeasibility.m, getShearletIdxs.m.
contains functions to simulate aligned reconstruction problems: decodeDISORDER.m, directionDISORDER.m, encodeDISORDER.m, errorFit.m, generateParametersExp.m, gradientDISORDER.m, precondDISORDER.m, pyramidDISORDER.m, SNRToLevels.m, synthesizeEncoding.m, synthesizeT.m, synthesizeY.m, Xsolver.m.
NOTE 1: Exemplary data is provided in the datasets exampleSpectrum.txt (used by fig0304.m), xGT.mat (used by fig0506.m), and GT.mat and Q[1-3].mat (used by fig0708.m). For runs without changing the paths, they should be placed in a folder
Data generated when running the scripts appears in subfolders Results5-6, Results7. Nifti files are generated in subfolder An-Ve with suffixes *_Aq.nii (no motion correction), *_Di.nii (with correction) and *_Re.nii (with robust correction).
NOTE 2: As for simulations, the provided execution should give a simplified version of the Figures in the paper, set 'quick=0' when calling fig0506.m to generate all the plots. As for reconstructions on real data fig0708.m these are computationally heavy; please, refer to the computation details in the manuscript to assess if your computing resources are adequate.
NOTE 3: The data to reproduce the experiments of Fig. 8 could not be linked to the release due to memory limitations. Please contact lucilio.cordero_grande@kcl.ac.uk if you are interested in these datasets (BSSFP.mat, FLAIR.mat, MPRAGE.mat, SPGR.mat, TSE.mat, used by fig0708.m with results in Results8-A and Results8-B).