1.2 Technical recommendations on clinical renal MRI


Aims

This task force will develop technical recommendations on how to measure renal MRI biomarkers on clinical scanners of 1.5T and 3T field strengths. If you want to start a clinical study, you will find here concrete guidelines on how to implement the acquisitions, and you can be reassured that the approach will be accepted by a broad body of experts in the field. 

The recommendations will be published open-access in a special issue of the journal MAGMA, detailing the process and rationale underlying the choices that were made. Detailed instructions on how to implement the recommendations will be made available as supplementary material in an open-access version-controlled repository. 

We expect this to lead to a significant improvement in efficacy of research as it eliminates the need for local development and optimisation. It will also form a driver for a more standardised approach, improving comparability of results by different teams, and establishing a common benchmark to evaluate future developments.

The recommendations will be developed by panels of experts. The panels are open to any MRI scientist with a relevant track record, and their composition can change dynamically. Ideally they cover experience with all MRI vendors and application areas.

The task force is led by Steven Sourbron (chair) and Pim Pullens (co-chair).


Approach

In the development of the recommendations the panels will use the following guiding principles:

Harmonisation by consensus: The panels will aim to identify a harmonised acquisition approach that will be defined using vendor-neutral terminology, and that constitutes, in the opinion of the panel, the optimal way of performing the measurement. Consensus should be driven by available scientific data and clinical results where available. If insufficient evidence is available, the panel will nevertheless aim to put forward a single recommendation, as this will establish a common reference for further development and reduce systematic differences in the literature. When a consensus cannot be achieved in particular areas, the panel will instead recommend (a set of) experiments that will help to resolve the question, and revisit the recommendation when the evidence becomes available. 

Guidance on implementation: The panel will complement the recommendations with guidance on how to implement them on clinical scanners. This will be in the form of detailed and complete lists of sequence parameters for specific machines of specific vendors and software versions, complemented where possible with files that can be installed directly. If the recommended approach is not available in particular machines of particular vendors, the panel may provide some guidance on how to set up acceptable alternative sequences.

Dynamic and version-controlled: The recommendations and their implementations will be version-controlled and revisited at regular intervals by the panel of experts, or ad-hoc when significant new evidence becomes available. They may be upgraded if the panel agrees that the new evidence is sufficiently strong to justify refining previous recommendations. 

Application-specific where needed: The optimal approach may depend on the clinical application (eg. native vs transplanted kidneys) and may also be biomarker specific (eg. an optimal DTI protocol for tractography may not be the same as one where only a cortical FA measurement is needed). Other applications may be limited by scan time, eg. when an extensive biomarker panel is needed. Where appropriate, each panel may therefore develop different recommendations for specific application areas.


In the longer term, the panels will also aim to develop materials and methods to qualify and rank recommended solutions and new proposals. Development of these materials and methods may be a longer-term process that is unlikely to be mature by the time the first versions of the recommendations are published.

Field testing: The panels will aim to supervise subsequent evaluation of the recommended protocols and collect reporting bias and repeatability coefficients for participating sites. This in combination with a meta-analysis, can help estimate statistical power for future clinical research studies by interested stakeholders (e.g. new sites or pharmaceutical companies).

Site qualification: We accept that exact implementation of recommended solutions does not guarantee compliance due to hardware differences. The panels will aim to provide alongside the recommendations also data for qualifying local implementations, such as example data on phantoms and healthy volunteers or repeatability data. 

Quality scores: We accept that ultimately the field has to move away from a largely consensus-driven approach to a fully evidence-based approach to developing recommendations. To prepare this transition the panels will work towards a scoring system to rank alternative solutions using quantitative quality metrics. 

Process

The recommendations will be developed for each renal imaging biomarker separately, going through the following stages:

Stage 1. Review of recent literature. Collect a complete list of papers reporting the biomarker in the past decade (or so).

Stage 2. Collection of protocols. Request authors to share the detailed technical specifications of the protocols that were used in previous studies and/or unpublished work where authors are willing to share technical details. Authorship of the contributions will be recorded so that due credit can be attributed. Protocols will not be made public without consent of the contributors.

Stage 3. Comparison of protocols. Review and tabulate protocols to identify key differences and similarities between different research groups, different vendors, and different models and software versions. 

Stage 4. Consensus formation. Review the output of the comparison in stage 3 with an aim to arrive at a consensus on a set of recommended protocols. The panels will use the Delphi method for consensus formation where appropriate to ensure opinions can be heard free from peer pressure. 

Stage 5. Publication. The recommendations will be published in a MAGMA special issue (deadline 1 july 2019) detailing the process and the rationale for the choices that were made. The detailed technical implementations will be published as supplementary material on a version-controlled open-access website such as github.

Stage 6. Maintenance. The panel will remain in existence though its membership may evolve. The open-access site will have a system to allow users to list issues, flag up novel developments and suggest changes. At regular intervals the panel will revisit the recommendations, review novel evidence and comments on the previous versions, and formulate an upgrade as appropriate.


Milestones

2018-07-01 Initiative launched with a call for panel members on the biomarkers reviewed in the special issue of NDT: T1 & T2, T2*, ADC & FA, ASL-RBF.

2018-09-01 Panels finalised and general guidelines refined and agreed, initiate call to authors for contribution of protocols.

2018-10-04 Prague plenary meeting: discussion and start of consensus formation, identification of areas of agreement and discrepancy.

2019-03-18 Aarhus meeting: finalise draft of recommendations and formally agree the final version.

2019-07-01 Submission of recommendations v1.0 for publication (hard deadline)


Expert panels - roles and responsibilities

The expert panels will be led by a chair (senior subject specialist) and a co-chair (junior subject specialist). The role of the chairs is to coordinate the process outlined above (eg. through regular TC's), ensure timely delivery, and act as lead authors on the accompanying publications. The role of the task force leads is to ensure consistency between the panels in the basic approach and the presentation of data and results. There are various ways that the other panel members can contribute to the process: 

  1. Contributing technical details of protocols for published or unpublished studies;
  2. Helping to collect, tabulate and compare protocols submitted;
  3. Helping to develop, organise, and document the online supplementary material (github or equivalent);
  4. Regularly taking part in teleconferences of the panel in order to define a consensus;
  5. Other substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work (please specify);
  6. Drafting the work;
  7. Revising the work critically for important intellectual content;
  8. Final approval of the version to be published;
  9. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Regarding authorship on the ensuing publications, we will adhere to the ICMJE criteria. In order to qualify for authorship a panel member will therefore need: at least one contribution from 1-5 AND at least one contribution from 6-7 AND 8 AND 9. Contributions that do not qualify for full authorship are welcome too and will be recognised in the acknowledgements section (eg. contributing protocols without taking part in the actual process of developing recommendations).


ASL Panel

  • Chair: Maria Seara (Pamplona, Spain)
  • Co-chair: Fabio Nery (London, UK): 1,2,3,4,6,7,8,9
  • Anita Harteveld (Utrecht, The Netherlands): 1,2,3,4,6,7,8,9.
  • Susan Francis (Nottingham, UK): 1,2,4,6,7,8,9
  • Xavier Golay (London, UK): 5 (major contribution in defining the approach to consensus formation), 7, 8, 9.
  • Alexandra Ljimani (Dusseldorf, Germany): 1,7,8,9
  • Katja Derlin (Hannover, Germany)
  • Charlotte Buchanan (Nottingham, UK): 1, 2, 4, 5 (acquisition), 6, 7, 8, 9.
  • Eleanor Cox (Nottingham, UK): 7, 8, 9
  • David Alsop (Boston, US): 1,2,5,7,8,9 with 5 representing working with Maria and Sue Francis on a multivendor comparison of FAIR and PCASL.
  • Ivan Pedrosa (Dallas, US): 1,4,6,8,9.
  • Ananth Madhuranthakam (Dallas, US): 1, 4, 6, 7, 8, 9.
  • Danny JJ Wang (USC, Los Angeles)
  • Agogho Odudu (Manchester, UK)
  • Philip Robson
  • Jeff Zhang (Utah, US): 1, 7, 8, 9
  • Kanishka Sharma (Leeds, UK): 1,2,4,8,9
  • Michael Gach (Washington University, St. Louis, USA)
  • Manuel Taso (Beth Israel Deaconess Medical Center, Boston, USA)
  • Pim Pullens (Gent, Belgium): 2,5 (task force setup),7,8,9
  • Steven Sourbron (Leeds, UK): 5 (task force setup, consistency with other panels), 7, 8, 9

BOLD Panel

  • Chair: Pottumarthi Prasad (Northshore, US)
  • Co-chair: TBC
  • Octavia Bane (Mount Sinai, US): 1,2,4,,6,7,8,9.
  • Menno Pruijm (Lausanne, Switzerland)
  • Lilach Lerman (Rochester, Minnesota, US): 4,7,8,9
  • Iosif Mendichovszky (Cambridge, UK): 2, 4, 6, 7, 8, 9.
  • Susan Francis (Nottingham, UK): 1,2,4,6,7,8,9
  • Stephen Textor (Rochester, US): 1,2,7,8,9.
  • Bastien Milani (Lausanne, Switzerland)
  • Sean Fain (Wisconsin, US): 1, 4, 7, 8, 9
  • Chunlei Liu (Berkeley, US): 1, 4, 6, 7, 8, 9
  • Kanishka Sharma (Leeds, UK): 1,2,4,8,9
  • Christoffer Laustsen (Aarhus, DK): 1,4,6,7,8,9
  • Thoralf Niendorf (Berlin, Ger): 1,4,7,8,9
  • Jeff Zhang (Utah, US): 1, 7, 8, 9
  • Pim Pullens (Gent, Belgium): 2,5 (task force setup),7,8,9
  • Steven Sourbron (Leeds, UK): 5 (task force setup, consistency with other panels), 7, 8, 9

DWI Panel

  • Chair: Eric Sigmund (New York, USA)
  • Co-chair: Alexandra Ljimani (Dusseldorf, Germany): 1,2,5 (data analysis and interpretation), 6,8,9
  • Nicolas Grenier (Bordeaux, France): 4,7,8,9
  • Iosif Mendichovszky (Cambridge, UK): 2, 4, 6, 7, 8, 9.
  • Jean-Paul Vallee (Geneva, Switzerland): 1, 4, 7, 8, 9
  • Anna Caroli (Bergamo, Italy): 1,4,7,8,9
  • Harriet Thoeny (Fribourg, Switzerland)
  • Susan Francis (Nottingham, UK): 1,2,4,6,7,8,9
  • Katja Derlin (Hannover, Germany)
  • Marcos Wolf (Vienna, Austria): 1, 4, 6, 7, 8, 9.
  • Martijn Froeling (Utrecht, The Netherlands): 1, 2, 5(acquisition analysis), 7, 8, 9
  • Kanishka Sharma (Leeds, UK): 1,2,4,8,9
  • Moritz Schneider (Munich, Germany): 1, 2, 4, 6, 7, 8, 9
  • Christoffer Laustsen (Aarhus, DK): 1,4,6,7,8,9
  • Thoralf Niendorf (Berlin, Ger): 1,4,7,8,9
  • Fabio Nery London, UK): 1,7,8,9
  • Pim Pullens (Gent, Belgium): 2,5 (task force setup),7,8,9
  • Steven Sourbron (Leeds, UK): 5 (task force setup, consistency with other panels), 7, 8, 9

T1&T2 Panel

  • Chair: Susan Francis (Nottingham, UK): 1,2,3,4,6,7,8,9.
  • Co-chair: Ilona Dekkers (Leiden, NL): 1,2,3,4,5,6,7,8,9
  • Marcos Wolf (Vienna, Austria): 1, 4, 6, 7, 8, 9.
  • David Buckley (Leeds, UK): 1,7,8,9
  • Jo Naish (Manchester, UK): 1,2,3,4,7,8,9.
  • Charlotte Buchanan (Nottingham, UK): 1, 2, 4, 5 (acquisition), 6, 7, 8, 9.
  • Eleanor Cox (Nottingham, UK): 1,2,4,6,7,8,9
  • Anneloes de Boer (Utrecht, The Netherlands): 1,2,3,4,6,7,8,9.
  • Kanishka Sharma (Leeds, UK): 1,2,4,8,9
  • Christoffer Laustsen (Aarhus, DK): 1,4,6,7,8,9
  • Matthew Robson (Oxford, UK)
  • Hildo Lamb (Leiden, NL): 1,4,6,7,8.
  • Pim Pullens (Gent, Belgium): 2,5 (task force setup),7,8,9
  • Steven Sourbron (Leeds, UK): 5 (task force setup, consistency with other panels), 7, 8, 9

Technical notes

Protocol collection (Pim Pullens)

Acquisition protocols will be collected centrally and harmonised to facilitate comparisons and generate a coherent structure and appearance between the outputs of the different panels. In order to facilitate this process we request the protocols are provided in the following way:

For all submissions: 

  • An ordered list of protocol authors for credit
  • A reference to the paper (or study) where the protocol is applied
  • Scanner type, software version and coil info

For Siemens:

  • PDF output of the protocol (mandatory)
  • If possible, edx/exar1 files
  • At least one DICOM file for each protocol step if possible

For Philips:

  • Text output of each sequence in the protocol (mandatory)
  • ExamCard if possible
  • At least one DICOM file for each protocol step if possible

For GE:

  • PDF file of the protocol (mandatory)
  • At least one DICOM file of each sequence in the protocol (mandatory)
  • Protocol Exchange files if possible
  • We really need DICOM and/or Protocol Exchange files, because the information in the PDF is very limited