Dry goods | fMRI standard reporting guidelines are fresh, come and increase your knowledge

Hello, Hello everyone
This is one brain cloud research circle , I'm Qiuqiu ~

as everyone knows , The replicability of research is crucial .

However, with the interdisciplinary and scientific development , Scientific articles are increasingly difficult to show all the details needed for accurate replication , It also brings a series of problems

Foreign researchers Poldrack And other people in 2008 Years published in NeuroImage A study in the journal outlines a set of reporting guidelines and list of assists for fMRI research methods and results .

that , How to publish an article effectively and accurately fMRI Related articles ？

Learn quickly ！

1

introduction

be engaged in fMRI Scholars of related work and research are widely distributed in Biology 、 psychology 、 Medical Engineering 、 Statistics and other different fields .

If the researcher is not accurate in the presentation of research details , It will not only make it difficult for other researchers and readers to understand , It will also affect the effective data mining and meta analysis of large-scale database metadata .

This article provides direct and effective guidelines , Outlines any fMRI What should be explained in detail in the paper , Designed to fMRI The method details and results of the paper are more complete .

In addition, the article finally puts forward a clearer list , Authors can use it to ensure that the paper contains all the necessary information presented in the guide .

meanwhile , The author points out that , This article should not be stereotyped , It should be seen as a starting point for encouraging debate , It aims to reach a set of widely accepted and flexible guidelines .

So , They set up a website ：

fmrimethods.org/.

On this website , Researchers can conduct discussions to ensure that they reflect current standards in the field and not just the opinions of a group of researchers .

2

Guide content

Indicate the range of subjects

The article should provide the basic demographic information of the research participants and other necessary information .

for example ,“ The subjects reported no history of mental or neurological diseases , And currently do not use any psychotropic drugs ”

If the subject sample is recruited in a targeted way , Then pay attention to the sampling strategy .

Besides , Pay attention to how many subjects are excluded from the study after data collection and the reasons for exclusion .

Indicate the tasks the subjects were asked to do and what they actually did

When describing fMRI Psychological tasks used in , You should try to provide enough details to ensure the reproducibility of the experiment , And this is often not easy .

Psychophysicists may be concerned about the perspective and brightness of stimuli , Economists may be concerned with the details of how to determine compensation .

Authors must use their judgment to decide which details are important for a particular study .

But in general , Rich details are better than too few details .

Clarify what is “Talairach Space ”

This term is easily confused in neuroimaging research , Therefore, researchers should be cautious when using it .

If the anterior commissure and posterior commissure are at the same level (AC–PC Line ) On , And the median plane contains this line , Then the brain or atlas is located in the talus space .

The brain is located in Talairach Space does not mean the existence of any particular brain shape or size , Nor does it mean that such a brain matches a specific template .

When it is only pointed out in the article that the coordinates are “Talairach Space ” in , Without more details , You should pay attention .

The author should specify the matching atlas or template or give the type of transformation used , And the specific details of spatial normalization methods such as the type of image being transformed .

This problem will directly affect the database work , Because database work needs to accurately map data to the public space of data sets generated by different methods .

Specify how to determine the region of interest

The region of interest ( region of interest,ROI ) It can be used to extract the estimated value of the induced signal , You can also limit the correction of multiple tests to a subset of all voxels .

In either case , Papers must describe how to determine ROI.

Of particular importance , Used for multiple test correction ROI ( Is often referred to as " Small volume correction ") It is determined independently of the specific test for calibration , Or use orthogonal comparison , Or scan independently .

If ROI It is anatomically determined , Then the anatomical boundary rules should be clearly specified if ROI It's a functional definition , So definition ROI The specific contrast used should be specified .

It is suggested that researchers request to provide in the corresponding format in the supplementary materials of the paper ROI Definition .

Provide enough detail to reproduce the analysis

FMRI analysis software package is very powerful , But it can also produce easily misunderstood results and even be misused .

Although at present most fMRI The study reports the use of general linear models ( GLM ) Analyze , But there are still substantial differences in how these models are specified and estimated .

Even based on GLM Under the framework of analysis , There are also many different modeling methods .

At present, most published studies use methods that are part of established software packages , And has been described in methodological publications .

However , It is not uncommon for papers to present results using a method that has not been previously described in methodological journals .

under these circumstances , The key is to describe the method in detail , So that others can reproduce .

The best reproducibility test is to allow others to reproduce the analysis directly on their own data .

Report statistical tests to support all claims

Any empirical claims of research should be supported by specific statistical tests .

Although it seems obvious , But this is a principle that is often violated in neuroimaging literature .

When activation is observed in a comparison , In another comparison, there is no activation , On this basis, it is concluded that there are differences between the two effects , This will happen, which is widespread in the literature and is Henson be called “ Image fallacy ” The situation of .

However , To prove that these two effects are different , A direct statistical comparison of these two effects is needed .

Again , The argument about differences must be supported by significant interactions across hemispheres or regions .

Authors should try to avoid suggesting that their activation area is the only area involved in the task .

If they really want to directly assert that one or more other areas are not active , This assertion should be accompanied by the size of the effect 、 Bayesian posterior probability of confidence interval or effect .

Describe and explain multiple test problems

fMRI Due to the high dimensionality of data, there will be a series of problems , But this is related to I The high risk of type a errors is related to the cost of many concurrent statistical tests .

therefore , It is necessary to clarify the scale of the multiple testing problem and how to deal with it .

The severity of the problem is described by the number of voxels tested and the smoothness of the data .

Examples of specific methods for multiple tests include voxel or cluster control family errors , Voxel control error detection rate , Or a formal heuristic that has been shown to control false positives in some objective way .

Be clear about the inferences that can be drawn from your method

for example , If you use uncorrected thresholds , Clearly state your unquantified control over family errors .

Corrected or both corrected and uncorrected inferences , It should be reported and clearly marked according to the type of correction .

When using cluster based inference , This should be clearly noted , And report the threshold used to create the cluster and the threshold of the cluster size .

Numbers and tables should exist independently

fMRI Effective presentation of results usually includes ： Number with threshold color coded statistical graph , A table listing significant activation locations .

The author's general advice is , The nature of the data presentation should follow the assumptions being tested .

therefore , If the hypothesis is tested at the group level , Then it may be more meaningful to present the group average graph , And research that tests assumptions about individual differences should present some data , Make these differences clear ( For example, scatter diagram or box diagram ).

Record quality control measures

stay fMRI There are many kinds of quality control measures applied in data collection and analysis , There is no set of general measures or methods across laboratories .

The author encourages the use of detailed literature quality control measures , In order to provide reviewers and readers with the best possible ability to estimate data or analysis with potential problems .

A special measurement method recommended by the author is the presentation of voxel masks used in group data analysis ( In supplementary materials or downloadable online format ), It demonstrates which voxels are included in the analysis .

3

List of standards

at present , Many other bioscience fields are currently engaged in a similar debate on the minimum information standards for methodological reporting , For example, microarray research MIAME Guidelines and koonsolt guidelines for clinical trials , The author hopes fMRI Work together to develop fMRI Community standards for method reporting .

In some areas such as clinical trials, a certain kind of list needs to be completed , In order to submit documents .

At the end of the article, a list of standards is presented , The website established by the author is also constantly improving and updating the list .

Last , If you are engaged in scientific research, it is necessary to formulate a list of research standards in the industry , And find this article useful , You might as well think about and practice every rule in detail , Enter the website to participate in the discussion .

reference

Poldrack, R. A. , Fletcher, P. C. , Henson, R. N. , Worsley, K. J. , Brett, M. , & Nichols, T. E. . (2008). Guidelines for reporting an fmri study. Neuroimage, 40(2), 409-414.

That's all for today's introduction

copy ： Autumn autumn
Typesetting ： Eucalyptus globulus
proofreading ：Uka Elder sister miaojun