{"paper_id":"3024727c-dd2d-4287-89c8-c0ef942d0fd1","body_text":"1 \n \nTitle \n \nqMRI-BIDS: an extension to the brain imaging data structure for quantitative magnetic \nresonance imaging data \n \nAuthors \nAgah Karakuzu1,2, Stefan Appelhoff3, Tibor Auer4, Mathieu Boudreau1,2; Franklin Feingold5, \nAli R. Khan6, Alberto Lazari7, Christopher J. Markiewicz5, Martijn J. Mulder8, Christophe \nPhillips9, Taylor Salo10, Nikola Stikov1,2, Kirstie Whitaker11 *, Gilles de Hollander12,13*  \n \nAffiliations \n1 NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal, Montréal, QC, \nCanada \n2 Montreal Heart Institute, Montreal, QC, Canada \n3 Center for Adaptive Rationality, Max Planck Institute for Human Development, Berlin, \nGermany \n4 NeuroModulation Lab, School of Psychology, University of Surrey, Guildford, UK \n5  Stanford University, Stanford, CA, USA \n6 Department of Medical Biophysics, Robarts Research Institute, University of Western \nOntario, London, Canada. \n7 Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical \nNeurosciences, University of Oxford  \n8 Department of Experimental Psychology, Utrecht University, Utrecht, the Netherlands \n9 GIGA Cyclotron Research Centre in vivo imaging, GIGA Institute, University of Liège, Liège, \nBelgium \n10 Florida International University, Miami, FL, USA \n11  The Alan Turing Institute, London, UK \n12 Zurich Center for Neuroeconomics (ZNE), Department of Economics, University of Zurich, \nZurich, Switzerland \n13  Spinoza Centre for Neuroimaging, Amsterdam, The Netherlands, \ncorresponding authors: Agah Karakuzu ( agahkarakuzu@gmail.com), Kirstie Whitaker \n(kwhitaker@turing.ac.uk), Gilles de Hollander (gilles.de.hollander@gmail.com)   \n \nAbstract \nThe Brain Imaging Data Structure (BIDS) established community consensus on the \norganization of data and metadata for several neuroimaging modalities. Traditionally, BIDS had \na strong focus on functional magnetic resonance imaging (MRI) datasets and lacked guidance \non how to store multimodal structural MRI datasets. Here, we present and describe the BIDS \nExtension Proposal 001 (BEP001), which adds a range of quantitative MRI (qMRI) applications \nto the BIDS application sphere. In general, the aim of qMRI is t o characterize brain \nmicrostructure by quantifying the physical MR parameters of the tissue via computational, \nbiophysical models. By proposing this new standard, we envision standardization of qMRI \nwhich makes multicenter dissemination of interoperable da ta possible. As a result, BIDS can \nact as a catalyst of convergence between qMRI methods development and application -driven \nneuroimaging studies that can help develop quantitative biomarkers for neural tissue \ncharacterization. Finally, our BIDS extension o ffers a common ground for developers to \nexchange novel imaging data and tools, reducing the practical barriers to standardization that is \ncurrently lacking in the field of neuroimaging. \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 \n \nIntroduction \nThe brain imaging data standard (BIDS) is an open source initiative from the neuroimaging \ncommunity that aids in standardizing neuroimaging data sets. BIDS was originally developed \nwith functional MRI (fMRI) applications in mind, describing experimental task blocks in \nrelation to a hierarchical organization of reconstructed MR images1. This convention engaged \nresearchers to share hundreds of open fMRI data on the openneuro platform 2,3 and develop \ninteroperable processing workflows that can seamlessly process these datasets 4. Popular \nexamples include the MRIQC5 and fmriprep6 pipelines, which can be executed even online for \nany valid BIDS fMRI dataset. Simila rly, the development of an MRI k -space data standard, \nISMRM-RD7, led open-source MRI reconstruction packages to adapt this convention and now \naids potential users in performing advanced reconstruction tasks with minimal effort 8,9. These \nsuccess stories from open science exemplify how data standards can change the landscape of \ncommunity-driven software for the better, leading to a collective change  in researchers’ \nbehaviour to adhere with FAIR (findability, accessibility, interoperability and reusability)  \nprinciples of scientific data10. Here we present our work extending the BIDS to include multi -\ncontrast MRI acquisitions. BIDS Extension Proposal 001 (BEP001) was merged into the \nstandard (on 23 February 2021) and focuses on quantitative MRI (qMRI) applications. \nQuantitative MRI methods map physical magnetic properties of the (brain) tissue. Their \napplication consists of  two steps: i) collecting multiple MRI images, where the contributions \nof effective micrometer-level MRI parameters is systematically manipulated by adapting very \nspecific acquisition parameters, and ii) fitting the resultant voxel intensity variations across the \nimages to a computational (biophysical) model 11. The results are a single or multiple \nquantitative map of the estimated parameters  across the imaged volume. The effective MRI \nparameters that are typically studied include longitudinal and transverse relaxation time \nconstants (T1 and T2, respectively), proton density (PD), magnetization transfer (MT), and local \ndiffusion coefficient (e.g., fractional anisotropy, FA, or mean diffusivity, MD). The multi -\nparametric mapping12 (MPM) protocol offers a set of acquisitions that can quantify more than \none MR parameter at a time. Another popular technique used in qMRI is field mapping, which \ncharacterizes inhomogeneities in MRI radiofrequency (RF) transmit (B1+) and receive (B1 -) \nprofiles, as well as static magnetic field (B0)  to correct qMRI parameter estimation errors due \nto these field inhomogeneities. \nThe earliest qMRI applications date back to the late 70’s 13 and primarily focused on \nrelaxometry, mapping of quantit ies such as T1 and T2* relaxation time. Since then, the field \nhas witnessed multiple waves of methods development, driven by technological advances and \nemerging trends in MRI research 14,15. Recently, with the surge of deep learning methods, the \ngamut of parameter estimation methods have become much larger than ever before  15-19. \nInterestingly, however, we still do not precisely know the healthy range of relaxation t ime \nvalues in a multi-center setting20 nor do we know how to establish diagnostically reliable tissue \ntyping protocols. This discrepancy highlights that multicenter standardization should be a \ncritical step toward evaluating the clinical potential of decades -long improvement s in the \nacquisition and processing of qMRI data. \nUnder more controlled research settings, qMRI offers obvious advantages over conventional \nMRI contrasts (e.g. T1 weighted images) in structural feature extraction. Given that MRI is not \na direct measurement of in vivo anatomical structures, voxel -wise morphometry analyses are \nsubjected to various biochemical and physiological confounders affecting the voxel \nintensity21.Hence, the capacity of disentangling MRI signal components lands qMRI as a more \nreliable approach to study structural variations 22. This makes qMRI particularly useful for \ncomparisons of the brain anatomy of differ ent (clinical) groups 23-25 and for more consistent, \nunbiased automated anatomical segmentation26-29. The same principle can be exploited to make \nqMRI sensitive to tissue microstructure, such as iron concentration or myelination. Recent meta \nanalyses revealed that a majority of qMRI methods are comparably sensitive to the myelin \ncontent30,31, although certain parameters such as myelin water fraction (MWF, relaxometry -\nbased) and macromolecular pool fraction (MPF, MT-based) appear to be more specific. \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n3 \n \nGiven the advantages offered by parametric maps in providing structural information and the \ncurrent landscape of myelin imaging methods, it seems likely that many more myelin imaging \nmethods leveraging the potential of qMRI will be developed in the future. This leads to one of \nour four main motivations behind covering qMRI methods in BIDS: to bring FAIR prin ciples \nto a variety of qMRI data that are finding widespread use in neuroimaging research. Other \nmotivations include i) driving open -source qMRI tools to adapt a consolidated input/output \nconvention, ii) creating standardized databases that can help simplify the use of qMRI in clinical \nand translational research, and iii) stimulating an open provision of qMRI data that can be \ncollected by imaging equipment that is available to a small group of researchers. \nDrawing upon the principles outlined in BIDS, we in troduce the first consensus data and \nmetadata organization standard for qMRI. This work is a culmination of years of effort (the \nearliest drafts of the BEP -001 date back to 2017) and discussion between neuroimaging \nresearchers and MRI methods developers around the globe. Our extension will not only aid in \norganizing qMRI data, but will also facilitate multi -center collaborative work, encourage \nneuroscientists to adapt advanced MR techniques and go a long way toward the standardization \nof qMRI methods. \n \nResults \nA new BIDS common principle: entity-linked file collections \nThe majority of qMRI methods necessitate the grouping of a set of similar images where \nspecific acquisition parameters are carefully varied. Furthermore, the images that are collected \nfor qMRI application do not usually have a clear \"weighting\" description (e.g., T1w, T2w), like \nconventional structural images. The novel concept of file collections decouples the semantics \nof logical group identification from contrast weighting labels or acqu isition sequence names \nthat are not originally developed for qMRI (e.g., FLASH). Instead, suffixes for such logical \nunits may indicate a generic MRI readout type (e.g., multi -echo gradient echo: MEGRE), a \nqMRI sequence name (e.g, magnetization prepared two rapid gradient echoes, MP2RAGE) or \na qMRI data collection framework (e.g., variable flip angle, VFA). Table-1 lists file collection \nsuffixes for various qMRI and fieldmap data, and the quantitative parameters they can derive. \nThese suffixes span a wide ra nge of qMRI applications including relaxometry, MT imaging, \nmultiparametric mapping, and RF field mapping. Application scope can be extended without \nnecessarily adding more suffixes. The BIDS qMRI appendix presents a set of rules and \nsuggestions to add new  qMRI suffixes to the specification  (https://bids-\nspecification.readthedocs.io/en/stable/99-appendices/11-qmri.html). \nNote that the use of file collections is no t exclusive to qMRI, anatomy imaging data, or even \nMRI. Any imaging modality calling for a file grouping logic to define a quantitative or \nqualitative application can benefit from this principle by specifying a descriptive suffix and \nfilename entity. Such changes would require additional BIDS extensions to create a valid file \ncollection.  \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n4 \n \n \nTable 1 - File collections of anatomy imaging data to derive parametric maps of longitudinal, \ntransverse and observed-transverse relaxation times (T1, T2 and T2*, respectively), proton density \n(PD), magnetization transfer ratio and saturation index (MTR and MTsat) and myelin water fraction \n(MWF). Relaxation rates (e.g., T1-1 and T2-1) and residual terms (e.g., M0) are excluded from the table \nfor brevity. \nqMRI application Suffix Derived \nmaps \nBIDS \nfolder \nReference \nMagnetization prepared \ntwo rapid gradient echoes \n(MP2RAGE) \nMP2RAGE T1 anat Marques et al. 201032 \nMultiparametric mapping \n(MPM) \nMPM T1, T2*, \nPD, MT \nanat Weiskopf et al. 201312 \nVariable flip angle (VFA) VFA T1, T2 anat Gupta et al. 197713 \nInversion recovery for T1 \nmapping (IRT1) \nIRT1 T1 anat Barral et al. 201033 \nMulti-echo spin-echo \n(MESE) \nMESE T2, MWF anat Carr and Purcell 195434, \nMackay et al. 199435 \nMulti-echo gradient-echo \n(MEGRE) \nMEGRE T2* anat Ma and Wehrli 199636 \nMagnetization transfer \nratio (MTR) \nMTR MT% anat Wolff et al. 198937 \nMagnetization transfer \nsaturation index (MTS) \nMTS MTsat anat Helms et al. 200838 \nDouble angle B1+ \nmapping \nTB1DAM B1+ fmap Insko and Bolinger 199339 \nB1+ mapping with 3D \necho-planar imaging \n(EPI) \nTB1EPI B1+ fmap Jiru and Klose 200640 \nActual flip angle imaging \n(AFI) \nTB1AFI B1+ fmap Yarnykh 200741 \nRapid B1+ mapping with \nTurboFLASH readout \nTB1TFL B1+ fmap Chung et al. 201042 \nSaturation-prepared with \n2 rapid gradient echoes \n(SA2RAGE) \nTB1SRGE B1+ fmap Eggenschwiler et al. \n201143 \nInter-scan motion \ncorrection using receive \nfield modulation \nRB1COR B1- fmap Papp et al. 201644 \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n5 \n \nTo distinguish individual files of a file collection, we introduced filename entities that are \nassociated with commonly altered acquisition parameters (e.g., flip angle) or with inherent \ncomponents of the same data (e.g., phase information), hence the name “entity -linked file \ncollection” (Table-2). \n \nTable-2 Filename entities representing an MRI acquisition parameter or designating an inherent part of \nthe reconstructed image (e.g., magnitude or phase). \nEntity format Entity values Associated acquisition \nparameter \nAssociated qMRI file \ncollections \necho-<index> 01,02,03,...,n EchoTime MEGRE, MESE, MPM \nflip-<index> 01,02,03,...,n  FlipAngle VFA, MTS, MPM \ninv-<index> 01,02,03,...,n  InversionTime IRT1, MP2RAGE \nmt-<label> on/off MTState MTR, MTS, MPM \npart-<label> mag/phase N/A MP2RAGE \n  \nIt is important to highlight that these entities cannot store acquisition parameter values in the \nfilename, but can only index or categorize them. Respective parameter values are stored in so-\ncalled \"sidecar JSON\"-files. Requirement level of these entities in relation to file collections are \npresented in the BIDS entity table appendix ( https://bids-\nspecification.readthedocs.io/en/stable/99-appendices/04-entity-table.html). \nData organization for qMRI file collections and quantitative parametric maps \nBy combining entities in the filename that represent different acquisition parameters (Table-2) \nwith entity -linked file collection suffixes (Table -1), BEP001 provides an intuitive way to \norganize filenames of most existing qMRI data. For example, raw data from MP2RAGE \nacquisitions comprises both magnitude and phase reconstructed images, acquired at two \nsuccessive inversion times (Fig -1a). The respective file collection for MP2RAGE (Fig -1c) \nclearly defines these components via part and inv-components, which are required for the \nMP2RAGE file collection. Note how the BIDS inheritance -rules do allow for using a single \nJSON-file to describe both phase- and magnitude-images, since these have identical acquisition \nparameters. In addition, the same collection suffix can be extended to specify its multi -echo \nvariant45 using the echo entity, which is made optional to MP2RAGE. For clarity, these specific \nuse cases are defined in the BIDS qMRI appendix. \nThe same logic applies to the raw images of double -angle B1+ mapping, identified by the \nTB1DAM suffix (Fig-1c). In this case, the maximum value of the flip entity indicates that the \ndata is collected over two flip angles. We recognize that an alternative approach to organize \nsuch data is stacking images at each flip angle into the 4th dimension of a Nifti-file, and storing \nthe corresponding metadata in vector form using a single JSONfile. This approach offers a less \ncrowded file list for this particular example. However, indexing acquisition parameter \ndependent variations across additional dimensions is less favorable for comprehensive qMRI \nmethods. For example, MPM collects raw data at different echo times, flip ang les and MT \npreparations with the option of phase reconstruction. After extended debates that took more \nthan a year, the qMRI -BIDS extension group ultimately concluded that this approach is less \nfavourable for human-readability of qMRI datasets, especially for multiparametric acquisition \nmethods where the number of images per protocol can go into the dozens. \n \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n6 \n \n \nFigure-1 a) Schematic of BIDS formatted raw quantitative MRI (qMRI) data representing MP2RAGE \n(anat) and TB1DAM (fmap) file collections, for which entity-linked metadata fields are highlighted for \nthe InversionTime (yellow), the FlipAngle (purple) and for the reconstructed image type (cyan). b) \nDerivatives of MP2RAGE and TB1DAM file collections generated by using pymp2rage and qMRLab \nto calculate T1 and B1+ maps, respectively, including a vendor-native derivative of UNIT1 images. c) \nFile organization of raw qMRI data for MP2RAGE and TB1DAM file collections, where respective \nlinking entities are highlighted for the inv entity (yellow, InversionTime), the flip entity (purple, \nFlipAngle) and the part entity (cyan, magnitude/phase). d) File organization of qMRI derivatives \nindicating how sidecar JSON files of quantitative maps generated by open-source software keeps a log \nof the input files (the BasedOn field) and associated acquisition parameters (FlipAngle in TB1map and \nInversionTime in B1map). \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n7 \n \nMetadata requirements for file collections and quantitative parametric maps \nFor the file collections, linking entities (Table -2) indicate a requirement for the respective \nacquisition parameters that are subject to change from image -to-image. Therefore, the entity \ntable appendix lists such parameters as required in relation to the corresponding file collection \nsuffix based on the descriptions made in the BIDS schema. Note that not all the parameters that \nchange across file collection images are captured by a linking entity, but may st ill be required \nfor data fitting. For example, the value of the FlipAngle parameter might (but does not \nnecessarily) covary with that of InversionTime between MP2RAGE file pairs; however, the \nfilenames are distinguished solely by the inv entity (since that is the crucial parameter that is \nswept over, whereas the flip angle could in principle remain the same). In addition, certain \nparameters that are constant across file collection images may be required as well. For example, \nRepetitionTimeExcitation and RepetitionTimePreparation are required metadata for an \nMP2RAGE acquisition. Such parameters are required when they are strictly necessary to \ncalculate the qMRI-maps that a specific acquisition scheme was designed to obtain; e.g., a T1-\nmap in case of MP2RAGE. BEP001 added an array of new metadata fields that may be required \nfor certain file collections (e.g. MTState, specifying whether an MT preparation is enabled in \nan MPM acquisition, associated with the mt linking entity) or provide supporting information \n(e.g., SpoilingRFPhaseIncrement, specifying the amount of incrementation applied to the phase \nof an excitation pulse). The complete list of metadata fields and their requirement levels for all \nthe qMRI file-collections are included in the BIDS release v1.5.0 and later. Currently, metadata \nconversions for some of these required fields have been implemented in dcm2niix 46, a \ncommonly used DICOM to NIfTI converter to create BIDS-compatible datasets. \nCertain quantitative parameters cannot be interpreted in absence of fundamental scanner \nspecifications. For example, to interpret relaxometry maps (e.g., T1map), the magnetic field \nstrength must be known. The BEP001 ensures that such requirements are met (again, see the \nqMRI Appendix in BIDS release v1.5.0 and later). Moreover, sidecar JSON files of quantitative \nmaps contain all the metadata values involved in the fitting by representing varying parameters \nin vector form and inheriting the constant ones from the raw images. To supplement the \nprovenance recording of parameter estimation process  with software -relevant details, the \nderived dataset and pipeline rules are respected as outlined in the modality agnostic files section \nof the main specification. \nFinally, the units and range of the fitted parameters have been standardized by BEP001 to define \ninterchangeable qMRI maps. For relaxometry -based parameters (e.g., T1map or T2map), the \ntime is described in seconds and the rate in reciprocal seconds or Hz. Wherever applicable, \nunitless ratio maps are described in percentage (e.g., MTRmap or MWFmap). For quantitative \nsusceptibility maps (i.e., Chimap) the local magnetic susceptibility is represented in parts per \nmillion. The RF transmit maps (i.e., TB1map) are specified in relative percentage units, where \n100% denotes the ideal case (i.e., measured flip angle equals the nominal value). Any deviations \nfrom 100% convey proportional deviations from the intended field strength. Please note that \ncertain quantitative parameters are described in arbitrary units, where the acceptable range of \nvalues vary based on the target anatomy (e.g., MTsat). \nCommunity software and the role of BIDS in standardizing qMRI \nAs of release v1.5.0, the BIDS validator can perform on BEP001-compatible qMRI data at the \ndirectory and filename level rules, based on the entity require ment levels specified per file \ncollection suffix. However, metadata-level validation rules have not been implemented yet. This \nis mainly because multi -vendor extraction of qMRI related metadata fields (e.g., MTState or \nRepetitionTimePreparation) is not sup ported by commonly used converters. Recently, we \nstarted working with dcm2niix 46 and BIDSme \n(https://github.com/CyclotronResearchCentre/bidsme) developers to identify and map vendor-\nspecific header information to BEP001-compatible metadata. \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n8 \n \n \nNevertheless, some metadata entities that are of profound importance to the accuracy of \nquantitative maps cannot be typically found in the vendor-native DICOM headers. For example, \nthe BIDS fields of RFSpoilingPhaseIncrement and SpoilingGradientMoment are two major \ndeterminants of T1 and B1+ estimation accuracy using spoiled gradient echo based \napplications47. Although this information is not provided by vendors, open -source pulse \nsequence development frameworks such as Pulse q48, PyPulseq 49, Gammasta r50, TOPPE 51, \nSequenceTree52, ODIN53 and RTHaw k54  can make a q MRI-tailored metadata annotation \npossible. An example implementation is qMRPullseq, a collection of publicly available vendor-\nneutral pulse sequences that are designed to export images in accordance with BEP001 format \nwithout hidden acquisition parameters 55. We highly encourage open -source MRI pulse \nsequence developers to use and contribute to the qMRI metadata annotations. This simple \nconsensus can remov e proprietary roadblocks from disseminating qMRI datasets  that \nincorporate key information on the reproducibility of data acquisition. \nMost qMRI methods can benefit from a plethora of BIDS applications 4 to prepare data for \nparameter estimation and downstream statistical analyses. There are several open-source tools \nemerging to perform qMRI fitting at multiple levels, like the hMRI -toolbox56, qMRLab 55, \nQUIT57, PyQMRI 58, QMRTools59, mrQ60, Madym 61, MITK -ModelFit 62, ROCKETSHIP 63, \nDCEMRI.jl64 and DCE@urLAB65. Giving these tools the ability to operate on BIDS formatted \ndata is an important step towards establishing interoperable qMRI processing pipelines. \nConclusion \nQuantitative MRI offers a rapidly developing set of techniques that can inform us about brain \n(micro)structure beyond what conventional MRI techniques have to offer66. We believe that, in \ncoming years, qMRI will become increasingly important to both clinical and fundamental brain \nscience. Therefore, a concrete standard for organizing and thereby also disseminating open \nqMRI data sets is much warranted. BEP001 extends the framework of the existing and very \nsuccessful BIDS standard, to develop a standard for qMRI in the form of a \"BIDS extension \nproposal\". To aid actual user adoption of this standard, it includes very precise descriptions of \nhow to use it in many real-life qMRI use-cases, as well as many example data sets. \nCurrently, obtaining qMRI data is still expensive and needs considerable expertise, which is not \nreadily available at many MRI facilities. Therefore, we also hope that BEP001 will aid \nresearchers that do not have easy access to such facilities to get familiar with qMRI data and \npotentially can even use open qMRI data sets for their particular research questions. \nFinally, the popularity of BIDS is likely in large part also due to some software packages that \nare designed around this standard and therefore extremely easy-to-use, when one's data adheres \nto the BIDS standard67. We hope that the success of BIDS in the domain of functional MRI will \nalso inspire and encourage MRI software developers to work on similar \"BIDS apps\" to make \nit easier to work with qMRI data, as well as make processing pipelines more open and \ntransparent. \nData Availability \nDuring and since the development of BIDS extension proposal 001, multiple data sets have been \nconverted to the new qMRI standard, in part also to stress -test the developing file naming \nschemes. Table-3 shows an (non-exhaustive) list of currently available qMRI data sets that are \nconverted to the extended BIDS standard (release 1.5.0).  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n9 \n \n \nTable-3 Various resources for example BIDS datasets making use of the specifications introduced by \nthe BEP001 extension proposal. \nName Description Link \nbids-examples A set of placeholder files for example \nqMRI dataset that are punctually \norganised according to BEP001 and \ntherefore a good reference for \ndevelopers and users. \nhttps://github.com/bids-\nstandard/bids-examples \nBEP001 \nexamples \nMP2RAGE, MP2RAGE-ME, MPM68, \nMTSAT, QSM, SA2RAGE and VFA \ndata collected and curated during the \nBEP001 development process. \nhttps://osf.io/k4bs5/ \nSpine generic   Multicenter MTS data for standardized \nquantitative imaging of human spinal \ncord69. \nhttps://spine-\ngeneric.readthedocs.io/ \nNeuromod Longitudinal data including MTS and \nMPRAGE, acquired from 6 individuals \nfor training artificial neural networks on \nhuman brain activity and behaviour70.  \nhttps://www.cneuromod.ca/ \nqMRPullseq Multicenter phantom data for \ncomparing the accuracy and \nreproducibility of MTS and TB1AFI \nacquisitions between vendor-specific \nand vendor-agnostic pulse sequence \nimplementations55. \nhttps://osf.io/5n3cu/ \nhMRI-toolbox   Example MPM dataset for in-vivo \nhistology using MRI (hMRI)56.  \nhttps://hmri-\ngroup.github.io/hMRI-\ntoolbox/ \n \nMethods \nCommunity-driven development of BEP001 \nThe development history of BEP001 spanned nearly 5 years. This extension was initiated \nfollowing mailing list discussions about  standardizing MP2RAGE32 datasets and including \nmulti-echo MRI acquisitions  in late 2016 ( https://bit.ly/bids_mailing). These discussions \nrevealed that BIDS was still lacking a generic convention for specifying structural acquisitions \nyielding multiple contrasts. In the summer of 2018, meeting were held to hear concerns and \nquestions from interested participants and to set an action plan for the development  during the \nannual INCF NeuroInformatics conference  in Montréal/Canada  \n(http://www.neuroinformatics2018.org/) and the OHBM meeting in Singapore \n(https://www.humanbrainmapping.org/i4a/pages/index.cfm?pageID=3821). As a first action, a \njoint-community meet ing was organized between MRI and neuroimaging scientists on 4 \nOctober 2018 ( https://www.ismrm.org/virtual-meetings/virtual-meetings-archive/) , where a \nconsensus decision was made on extending the specification for a variety of qMRI methods . \nAfter this meeting, a standard operational procedure was established and followed to advance \nthe proposal, focusing on both transparency and accessibility to other researchers (Fig. 2).  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n10 \n \n \nFigure-2 Summary of the standard operational procedure for improving BEP001. Outcomes from the \nmonthly meetings (a) are transferred to a central GitHub repository, opened for more elaborate public \ndiscussions via issues and merged into the proposal through peer-reviewed pull requests (b). BEP001 is \ninclusive to all communities who would like to contribute to the proposal or keep themselves up-to-date \nwith the latest developments. \n \nInterim outcomes from the development were presented in the 2020 annual conferences of \nOHBM71 and ISMRM 72 to reach out more neuroimaging and MRI  physics researchers, \nrespectively. Following another year of development on the specification, example datasets and \napplications, BIDS incorporated and released BEP001 as part of their version 1.5.0. The main \nproblems identified and resolved during the development are outlined in the following section, \nlaying out the methodology of how qMRI can be incorporated into BIDS. \nExtending an existing standard for new use cases \nBIDS traditionally focused on conventional anatomical images that are collected in functional \nMRI experiments and whose contrast characteristics are well-defined (i.e., mostly T1-weighted \nimages). This posed a challenge for the naming scheme of collections of multimodal images \nused in qMRI. Unlike conventional structural data, qMRI inputs are usually formed by \ncollections of images where specific acquisition parameters are systematically manipul ated. \nMoreover, the line separating contrast characteristics between these images is blurred. A \nconcrete example: in a multi-echo GRE acquisition with a long TRs, early echoes will be mostly \nPD- and B1+/B1- signal-weighted, whereas later echoes will be inc reasingly T2*-weighted. \nMost echoes will show a contrast that is the result of a mixture of underlying physical properties. \nThis ambiguity disqualifies MRI weightings (e.g., T1w or T2starw) as suffix labels to specify \ninterchangeable qMRI datasets. The use  of often proprietary acquisition sequence names like \n\"FLASH'' (fast low angle shot) or \"GRE\" (gradient-recalled echo) as a suffix turned out to also \nbe undesirable, because different MRI vendors use different naming conventions and, \nmoreover, one type of sequence can often be used for numerous qMRI applications. To address \nthis problem, BEP001 introduced a new common principle: file collections. \nA second challenge that BEP001 addressed pertains to standardizing the data organisation of \nquantitative parametric maps. One central challenge of such maps is that the calculations on \nwhich they are based can be made both by proprietary vendor software run on the scanner \nsystem, or offline using open -source workflows. The resultant map can be de scribed as \nderivative data in either case, yet the former lacks provenance of the whole calculation process \nand may not export the raw inputs to the calculation.  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n11 \n \nAcknowledgements \nThe autho rs would like to acknowledge the work by other contributors to BIDS, and in \nparticular those that contributed to BEP-001 via the Github repository, intermediate meetings, \nas well as a first draft on Google Drive. For BEP-001, recorded contributions include those from \nSuyash Bhogawar, Julien Cohen -Adad, Elizabeth Dupre, Chris Gorgolewski , Daniel \nHandwerker, Michael Harms, Ilana Leppert, Tobias Leutritz, Dylan Nielson, Julien Sein, Isla \nStaden, Wietske van der Zwaag, and Tobias Wood.  \nThis research was funded in part by the Wellcome Trust [Grant 109062/Z/15/Z to AL]. For the \npurpose of Open Access, the author has applied a CC BY public copyright licence to any Author \nAccepted Manuscript version arising from this submission.  \nTA’s work has been funded by the Biotechnology and Biological Sciences Research Council, \nLondon (BB/S008314/1). \nC.P. is supported by the F.R.S.-FNRS, Belgium.  \nG.H. was funded by a Rubicon grant from the Dutch Research Council (NWO).  \nA.K. is supported by  Canada First Research Excellence Fund through the TransMedTech \nInstitute, Canadian Open Neuroscience Platform (CONP) and International Society for \nMagnetic Resonance in Medicine (ISMRM).  \n \nAuthor contributions \nA.K., G.H. and K.W. prepared the original manuscript ; A.K., G.H. and K.W. developed the \ninitial draft of the standard and managed community contributions. A.K. merged the extension \nproposal to the main BIDS specification. G.H. and K.W. supervised the project . A.K., S.A, \nT.A., M.B ., F.F., A.K., A.L., C.M., M.M., C.P., T.S., N.S., K.W. and G.H. contributed to \nmeetings and drafts outlining the extension proposal. A.K., S.A, T.A., M.B., F.F., A.K., A.L., \nC.M., M.M., C.P., T.S., N.S., K.W. and G.H. revised the original manuscript.  \n \nReferences \n \n1 Gorgolewski, K. J. et al. The brain imaging data structure, a format for organizing and \ndescribing outputs of neuroimaging experiments. Scientific data 3, 1-9 (2016). \n2 Markiewicz, C. J.  et al. OpenNeuro: An open resource fo r sharing of neuroimaging \ndata. bioRxiv, 2021.2006.2028.450168, doi:10.1101/2021.06.28.450168 (2021). \n3 Poldrack, R.  et al.  Toward open sharing of task -based fMRI data: the OpenfMRI \nproject. Frontiers in Neuroinformatics 7, 12 (2013). \n4 Gorgolewski, K. J.  et al.  BIDS apps: Improving ease of use, accessibility, and \nreproducibility of neuroimaging data analysis methods. PLoS computational biology  \n13, e1005209 (2017). \n5 Esteban, O. et al. MRIQC: Advancing the automatic prediction of image quality in MRI \nfrom unseen sites. PloS one 12, e0184661 (2017). \n6 Esteban, O. et al. fMRIPrep: a robust preprocessing pipeline for functional MRI. Nature \nmethods 16, 111-116 (2019). \n7 Inati, S. J. et al. ISMRM Raw data format: A proposed standard for MRI raw datasets. \nMagnetic resonance in medicine 77, 411-421 (2017). \n8 Hansen, M. S. & Sørensen, T. S. Gadgetron: an open source framework for medical \nimage reconstruction. Magnetic resonance in medicine 69, 1768-1776 (2013). \n9 Maier, O. et al. CG‐SENSE revisited: Results from the first ISMRM reproducibility \nchallenge. Magnetic resonance in medicine 85, 1821-1839 (2021). \n10 Wilkinson, M. D. et al. The FAIR Guiding Principles for scientific data management \nand stewardship. Scientific data 3, 1-9 (2016). \n11 Novikov, D. S., Kiselev, V. G. & Jespersen, S. N. On modeling. Magn Reson Med 79, \n3172-3193, doi:10.1002/mrm.27101 (2018). \n12 Weiskopf, N. et al. Quantitative multi-parameter mapping of R1, PD(*), MT, and R2(*) \nat 3T: a multi -center validation. Front Neurosci 7, 95, doi:10.3389/fnins.2013.00095 \n(2013). \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n12 \n \n13 Gupta, R. K. A new look at the method of variable nutation angle for the measurement \nof spin -lattice relaxation times using fourier transform NMR. Journal of Magnetic \nResonance (1969)  25, 231 -235, doi: https://doi.org/10.1016/0022-2364(77)90138-X \n(1977). \n14 Stikov, N., Trzasko, J. D. & Bernstein, M. A. Reproducibility and the future of MRI \nresearch. Magnetic Resonance in Medicine  82, 1981 -1983, \ndoi:https://doi.org/10.1002/mrm.27939 (2019). \n15 Lundervold, A. S. & Lundervold, A. An overview of deep learning in medical imaging \nfocusing on MRI. Zeitschrift für Medizinische Physik 29, 102-127 (2019). \n16 Golkov, V. et al. Q-space deep learning: twelve-fold shorter and model-free diffusion \nMRI scans. IEEE transactions on medical imaging 35, 1344-1351 (2016). \n17 Yoo, Y. et al. Deep learning of joint myelin and T1w MRI features in normal-appearing \nbrain tissue to distinguish between multiple sclerosis patients and healthy controls. \nNeuroImage: Clinical 17, 169-178 (2018). \n18 Lyu, Q. & Wang, G. Quantitative MRI: absolute T1, T2 and proton density parameters \nfrom deep learning. arXiv preprint arXiv:1806.07453 (2018). \n19 Wu, Y., Ma, Y., Du, J. & Xing, L. Accelerating quantitative MR imaging with the \nincorporation of B1 compensation using deep learning. Magnetic Resonance Imaging \n72, 78-86 (2020). \n20 Bojorquez, J. Z. et al. What are normal relaxation times of tissues at 3 T? Magn Reson \nImaging 35, 69-80, doi:10.1016/j.mri.2016.08.021 (2017). \n21 Weinberger, D. R. & Radulescu, E. Finding the Elusive Psychiatric “Lesion” With 21st-\nCentury Neuroanatomy: A Note of Caution. American Journal of Psychiatry 173, 27-\n33, doi:10.1176/appi.ajp.2015.15060753 (2015). \n22 Lorio, S.  et al.  Neurobiological origin of spurious brain morphological changes: A \nquantitative MRI study. Human brain mapping 37, 1801-1815 (2016). \n23 Draganski, B. et al. Regional specificity of MRI contrast parameter changes in normal \nageing revealed by voxel -based quantification (VBQ). Neuroimage 55, 1423 -1434 \n(2011). \n24 Lommers, E. et al. Voxel-Based quantitative MRI reveals spatial patterns of grey matter \nalteration in multiple sclerosis. Human Brain Mapping  42, 1003 -1012, \ndoi:https://doi.org/10.1002/hbm.25274 (2021). \n25 Weiskopf, N., Mohammadi, S., Lutti, A. & Callaghan, M. F. Advances in MRI -based \ncomputational neuroanatomy: from morphometry to in-vivo histology. Current opinion \nin neurology 28, 313-322 (2015). \n26 Weiskopf, N., Callaghan, M. F., Josephs, O., Lutti, A. & Mohammadi, S. Estimating \nthe apparent transverse relaxation time (R2*) from images with different contrasts \n(ESTATICS) reduces motion artifacts. Frontiers in neuroscience 8, 278 (2014). \n27 Lutti, A., Dick, F., Sereno, M. I. & Weiskopf, N. Using high -resolution quantitative \nmapping of R1 as an index of cortical myelination. Neuroimage 93, 176-188 (2014). \n28 Haast, R. A. M., Ivanov, D., Formisano, E. & Uludaǧ, K. Reproducibility and reliability \nof quantitative and weighted T1 and T2 ∗ mapping for myelin -based cortical \nparcellation at 7 Tesla. Frontiers in neuroanatomy 10, 112 (2016). \n29 Dinse, J. et al. A cytoarchitecture-driven myelin model reveals area-specific signatures \nin human primary and secondary areas using ultra -high resolution in-vivo brain MRI. \nNeuroimage 114, 71-87 (2015). \n30 Mancini, M. et al. An interactive meta-analysis of MRI biomarkers of myelin. eLife 9, \ndoi:10.7554/elife.61523 (2020). \n31 Lazari, A. & Lipp, I. Can MRI measure myelin? Systematic review, qualitative \nassessment, and meta -analysis of studies validating microstructural imaging with \nmyelin histology. Neuroimage, 117744 (2021). \n32 Marques, J. P.  et al.  MP2RAGE, a self bias -field corrected sequence for improved \nsegmentation and T1 -mapping at high field. Neuroimage 49, 1271 -1281, \ndoi:10.1016/j.neuroimage.2009.10.002 (2010). \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n13 \n \n33 Barral, J. K. et al. A robust methodology for in vivo T1 mapping. Magnetic resonance \nin medicine 64, 1057-1067 (2010). \n34 Carr, H. Y. & Purcell, E. M. Effects of diffusion on free precession in nuclear magnetic \nresonance experiments. Physical review 94, 630 (1954). \n35 MacKay, A. et al. In vivo visualization of myelin water in brain by magnetic resonance. \nMagn Reson Med 31, 673-677 (1994). \n36 Ma, J. & Wehrli, F. W. Method for image -based measurement of the reversible and \nirreversible contribution to the transverse -relaxation rate. Journal of Magnetic \nResonance, Series B 111, 61-69 (1996). \n37 Wolff, S. D. & Balaban, R. S. Magnetization transfer contrast (MTC) and tissue water \nproton relaxation in vivo. Magnetic resonance in medicine 10, 135-144 (1989). \n38 Helms, G., Dathe, H. & Dechent, P. Quantitative FLASH MRI at 3T using a rational \napproximation of the Ernst equation. Magn Res on Med  59, 667 -672, \ndoi:10.1002/mrm.21542 (2008). \n39 Insko, E. K. & Bolinger, L. Mapping of the radiofrequency field. Journal of Magnetic \nResonance, Series A 103, 82-85 (1993). \n40 Jiru, F. & Klose, U. Fast 3D radiofrequency field mapping using echo-planar imaging. \nMagnetic Resonance in Medicine  56, 1375 -1379, \ndoi:https://doi.org/10.1002/mrm.21083 (2006). \n41 Yarnykh, V. L. Actual flip-angle imaging in the pulsed steady state: a method for rapid \nthree-dimensional mapping of the transmitted radiofrequency field. Magn Reson Med \n57, 192-200 (2007). \n42 Chung, S., Kim, D., Breton, E. & Axel, L. Rapid B1+ mapping using a preconditioning \nRF pulse with TurboFLASH readout. Magnetic resonance in medicine  64, 439-446 \n(2010). \n43 Eggenschwiler, F., Kober, T., Magill, A. W., Gruetter, R. & Marques, J. P. SA2RAGE: \nA new sequence for fast B1+‐mapping. Magnetic resonance in medicine 67, 1609-1619 \n(2012). \n44 Papp, D., Callaghan, M. F., Meyer, H., Buckley, C. & Weiskopf, N. Correction of inter‐\nscan motion artifacts in quantitative R1 mapping by accounting for receive coil \nsensitivity effects. Magnetic resonance in medicine 76, 1478-1485 (2016). \n45 Caan, M. W. A. et al. MP2RAGEME: T1, T2*, and QSM mapping in one sequence at \n7 tesla. Human brain mapping 40, 1786-1798 (2019). \n46 Li, X., Morgan, P. S., Ashburner, J., Smith, J. & Rorden, C. The first step for \nneuroimaging data analysis: DICOM to NIfTI conversion. Journal of neuroscience \nmethods 264, 47-56 (2016). \n47 Yarnykh, V. L. Optimal radiofrequency and gradient spoiling for improved accuracy of \nT1 and B1 measurements using fast steady -state techniques. Magnetic Resonance in \nMedicine 63, 1610-1626, doi:10.1002/mrm.22394 (2010). \n48 Layton, K. J.  et al.  Pulseq: a rapid and hardware‐independent pulse sequence \nprototyping framework. Magnetic resonance in medicine 77, 1544-1552 (2017). \n49 Ravi, K. S., Geethanath, S. & Vaughan, J. T. PyPulseq: A python package for mri pulse \nsequence design. Journal of Open Source Software 4, 1725 (2019). \n50 Cordes, C., Konstandin, S., Porter, D. & Günther, M. Portable and platform‐\nindependent MR pulse sequence programs. Magnetic resonance in medicine 83, 1277-\n1290 (2020). \n51 Nielsen, J. F. & Noll, D. C. TOPPE: A framework for rapid prototyping of MR pulse \nsequences. Magnetic resonance in medicine 79, 3128-3134 (2018). \n52 Magland, J. F., Li, C., Langham, M. C. & Wehrli, F. W. Pulse sequence programming \nin a dynamic visual environment: SequenceTree. Magnetic resonance in medicine 75, \n257-265 (2016). \n53 Jochimsen, T. H. & Von Mengershausen, M. ODIN —object-oriented development \ninterface for NMR. Journal of Magnetic Resonance 170, 67-78 (2004). \n54 Santos, J. M., Wright, G. A. & Pauly, J. M.   1048-1051 (IEEE). \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint \n\n14 \n \n55 Karakuzu, A. et al. qMRLab: Quantitative MRI analysis, under one umbrella. Journal \nof Open Source Software 5, 2343 (2020). \n56 Tabelow, K. et al. hMRI–A toolbox for quantitative MRI in neuroscience and clinical \nresearch. Neuroimage 194, 191-210 (2019). \n57 C Wood, T. QUIT: QUantitative imaging tools. Journal of Open Source Software  3, \n656 (2018). \n58 Maier, O., Spann, S. M., Bödenler, M. & Stollberge r, R. PyQMRI: an accelerated \nPython based quantitative MRI toolbox. Journal of Open Source Software  5, 2727 \n(2020). \n59 Froeling, M. QMRTools: a Mathematica toolbox for quantitative MRI analysis. Journal \nof Open Source Software 4, 1204 (2019). \n60 Mezer, A. et al. Quantifying the local tissue volume and composition in individual \nbrains with magnetic resonance imaging. Nature medicine 19, 1667-1672 (2013). \n61 Berks, M., m Parker, G. J., Little, R. & Cheung, S. Madym: A C++ toolkit for \nquantitative DCE-MRI analysis. Journal of Open Source Software 6, 3523 (2021). \n62 Debus, C. et al. MITK-ModelFit: A generic open-source framework for model fits and \ntheir exploration in medical imaging –design, implementation and application on the \nexample of DCE-MRI. BMC bioinformatics 20, 1-18 (2019). \n63 Barnes, S. R.  et al.  ROCKETSHIP: a flexible and modular software tool for the \nplanning, processing and analysis of dynamic MRI studies. BMC medical imaging 15, \n1-20 (2015). \n64 Smith, D. S., Li, X., Arlinghaus, L. R., Yankeelov, T. E. & Welch, E. B. DCEMRI. jl: \na fast, validated, open source toolkit for dynamic contrast enhanced MRI analysis. \nPeerJ 3, e909 (2015). \n65 Ortuño, J. E. et al. DCE@ urLAB: a dynamic contrast-enhanced MRI pharmacokinetic \nanalysis tool for preclinical data. BMC bioinformatics 14, 1-17 (2013). \n66 Weiskopf, N., Edwards, L. J., Helms, G., Mohammadi, S. & Kirilina, E. Quantitative \nmagnetic resonance imaging of brain anatomy and in vivo histology. Nature Reviews \nPhysics 3, 570-588, doi:10.1038/s42254-021-00326-1 (2021). \n67 Yarkoni, T. et al. PyBIDS: Python tools for BIDS datasets. Journal of open source \nsoftware 4 (2019). \n68 Callaghan, M. F. et al. Example dataset for the hMRI toolbox. Data in brief 25, 104132 \n(2019). \n69 Cohen-Adad, J.  et al.  Open-access quantitative MRI data of the spinal cord and \nreproducibility across participants, sites and manufacturers. Sci Data  8, 219, \ndoi:10.1038/s41597-021-00941-8 (2021). \n70 Boyle, J. et al. in 26th Annual Meeting of the Organization for Human Brain Mapping. \n71 Hollander, G.  et al.  in 26th Annual Meeting of the Organization for Human Brain \nMapping.   (OHBM). \n72 Karakuzu, A. et al. in Proceedings of ISMRM 28th Annual Meeting.   (ISMRM). \n \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 28, 2021. ; https://doi.org/10.1101/2021.10.22.21265382doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}