2025 Registered Abstracts

2025 ISMRM Annual Meeting
Registered Abstracts

Instructions for Authors

What is a registered abstract?

A Registered Abstract is an abstract where research methods are submitted and assessed by reviewers before the results are known.

Examples:

1. A newly published CNN model is applied to a new dataset. The registered abstract should contain a clearly defined research question (e.g. does this model outperform previous gold standards, e.g. U-Net) and study design (e.g. training on N-subjects chosen pseudo-randomly, then verified on M-subjects and tested on K-subjects, using PSNR as quality metric, etc.)
2. A new method for reducing noise in the measurements on a low field scanner was presented at last year’s ISMRM and is now applied to a different low-field scanner. The registered abstract should contain a clearly defined research question (e.g. does this method work on the other scanner model?) and study design (e.g. how the method needs to be modified if it needs to be modified at all; how much the SNR has to improve to be considered a success; how many times the noise measurement needs to be taken for statistically useful results, etc.)
3. A researcher has developed a new method for reconstructing MRF data that they think is faster than and of equal quality to standard methods. The registered abstract should contain a clearly defined research question (e.g. is my method faster than and of equal quality to standard methods, e.g. low-rank subspace reconstruction implemented in BART?) and study design (e.g. what hardware should the methods be compared on, what data should the methods be compared on, how are these choices justified, which metric will be used to assess quality of results, etc.)
4. A new sequence was published a few years ago that indicates stroke’s onset time, this is now used clinically and a researcher is doing a clinical trial to determine the accuracy of the method. The registered abstract should contain a clearly defined research question (e.g. what is the uncertainty in stroke onset estimation using the new sequence?) and study design (have data been collected but not yet analysed or they need to be collected, full analysis pipeline including data exclusion criteria, sample size justification, sources of uncertainty, sites harmonisation in case of multiple site study, etc.)

What can and cannot be submitted as registered abstract

Almost anyone can submit a registered abstract! This format is open for submissions in any of the standard submission categories. Registered abstracts are for research that is attempting to validate a carefully defined hypothesis. We especially encourage researchers who want to validate/translate preliminary work (e.g., a previous ISMRM abstract) to submit in this category.

Why should I submit a registered abstract?

Pre-registering your research shows credibility and a dedication to best practices. Submitting research in this format has the potential to reduce the amount of false positive results presented at the ISMRM meeting while fostering the translation towards clinical application and more robust research tools. Researchers that have pre-registered their work also report that it helps with clarity of thought, experiment design, and staying on track with the original research question. More info about registrations here: https://help.osf.io/article/603-benefits-of-registration

How is the submission process different from conventional abstracts?

When submitting a registered abstract you need to click “Submit a registered abstract” on the submission page. This will take you to a submission form for registered abstracts that differ from the conventional abstracts in three key ways:

1. The registered abstracts should only include an Introduction and Methods section. The Results and Discussion sections will be submitted in April, in time for the annual meeting. Only the first submission (Introduction and Methods) is peer-reviewed.
2. The total word count shall not exceed 500 words (unlike conventional abstracts that can be up to 750 words long).
3. Only two figures/tables can be submitted.

How will Registered Abstracts be scored?

Just like for conventional abstracts, reviewers are asked to score the abstracts on three qualities. For conventional abstracts these qualities are: impact, quality and novelty. For registered abstracts the qualities are: impact, quality, and transparency (see instructions for reviewers for further details). In addition the reviewers will confirm that certain minimum requirements for a registered abstract have been met: 1. Does the abstract contain an explicit testable hypothesis/research question? 2. Have the authors shown that they can feasibly acquire the necessary data to present results in time for the annual meeting? If certain circumstances make it particularly easy to collect data that would seem hard for others, authors are encouraged to include this in the abstract to allow reviewers to see the feasibility of the experiments.

What should be included in a registered abstract?

There are many ways to write a pre-registration, which is what a registered abstract really is, but the most important component is a carefully prepared methods section with an explicit hypothesis and data analysis plan. The introduction should motivate the study (why should anyone care about this result?). The Methods section should include an appropriate statistical plan (e.g., sample size calculation, effect size estimation, etc.) Pre-registration templates that outline core details to include and record in the pre-registration can be found on e.g., OSF (https://docs.google.com/docu…) and asPredicted.org (https://aspredicted.org/kv692.pdf).

Presentation

We expect that these abstracts will be presented as Power Pitch or Digital posters, which will allow presenters to submit results ahead of the actual meeting, as well as allowing for excellent discussion between audience and presenters. The results and discussion sections (250 words and up to 3 figures) must be submitted on or about 31 March 2024; without submission, the abstract will be withdrawn.

Instructions for Reviewers

What is a registered abstract?

A Registered Abstract is an abstract where research methods are submitted and assessed by reviewers before the results are known.

How is a registered abstract reviewed?

Each registered abstract will be reviewed by reviewers from the primary submission category as well as the reproducible research – registered abstract category.

The abstract is reviewed on the same scale as conventional abstracts (1-7) with 1 being the highest score and 7 the lowest score. The three dimensions that the registered abstracts should be reviewed on are: impact, quality, and transparency. Note that this is different from conventional abstracts that score novelty rather than transparency. Note that a high score on one of the axes does not necessarily mean the abstract gets a high score on the others. Abstracts can, for example, have high quality and low impact or high transparency and low quality, etc.

In addition to these scores, registered abstracts have to achieve two minimum requirements for acceptance: 1) The abstract has to include an explicit testable hypothesis. 2) The abstract has to show feasibility to gather and analyze data in time for the annual meeting.

Below follow further details on how to score the abstracts on each different criteria.

How to score a registered abstract on impact

For registered abstracts, impact does not come from impressive results (as results are not yet known), rather on the importance of answering the research question. For example, if the answer to “is method A more motion robust than method B?” would allow for clinical translation of the more motion robust method, that could warrant a high impact score. Conversely, an abstract regarding fine tuning of a parameter for a method that is far from clinical translation would have low impact. Impact could also come from the ability to definitively answer a question. For example, a large well powered study on the effectiveness of some method would yield a higher impact score than a smaller one. In summary, impact is the degree to which the work will influence the field, including how likely the work is to change scientific discovery, clinical practice or further technique development and capability.

How to score a registered abstract on quality

The quality score should reflect both the quality of the research (including study design, appropriateness of the research goal/question, rigor of the statistical & data analysis) AND the clarity with which the work is presented (easy to understand and well written text as well as clear and well labeled figures).

How to score a registered abstract on transparency

Registered abstracts are scored on transparency, as transparency is a key requirement for reproducibility. In this context, transparency means that the experiments and data analyses are well defined (all relevant conditions and variables clearly listed) to reduce researcher degrees of freedom after data has been collected. Transparency can also mean that reasons for design choices are justified (e.g., based on pilot data or previous literature). Finally, transparency can be scored on the use of other good open science practices, such as openly shared code and data.

How to check a registered abstract for minimum requirements

Reviewers of registered abstracts are asked to confirm whether minimum requirements have been met.

For a registered abstract to be accepted, a majority of reviewers need to be able to identify an explicit hypothesis. The reviewer will have to answer yes/no to the question “does the abstract contain an explicit testable hypothesis?” and if the reviewer answers yes, they will be asked to state what the hypothesis is (one sentence).

Secondly, to avoid accepting abstracts that fail to submit results in time for the meeting, reviewers should also answer the yes/no/unsure question “Is it possible to gather the required data to finish this study in time for the annual meeting?” More reviewers need to answer yes than no to this question for acceptance. When making this judgment, reviewers should consider the proposed experiments and whether data collection has started (and the data is just not yet analyzed) and whether the data collection is easy (e.g. phantom data) or hard (e.g. specific patient populations).