A proposal for a system that facilitates quantitative replication of fMRI studies

Stand-By Time

Tuesday, June 27, 2017: 12:45 PM  - 2:45 PM 

Submission No:

1547 

Submission Type:

Abstract Submission 

On Display:

Monday, June 26 & Tuesday, June 27 

Authors:

Johan Jansma1, Geert-Jan Rutten1

Institutions:

1Elisabeth-TweeSteden Hospital, Tilburg, Netherlands

First Author:

J. Jansma    -  Lecture Information | Contact Me
Elisabeth-TweeSteden Hospital
Tilburg, Netherlands

Introduction:

In many fMRI studies, individual brain activity maps are averaged in order to test hypotheses at group-level. For this purpose, individual scans are spatially normalized to a brain model in a standard data space, often MNI space. Using this method, an informative group map can be created by averaging signals for each voxel separately. It has become common practice to test hypothesis in each voxel separately, using a mass univariate testing approach. Because of the high number of results that are generated (i.e. the number of tested voxels), test results are communicated in graphical form, and each voxel that shows a significant signal change is displayed on top of a standard anatomy. This visual illustration is typically combined with a table providing the MNI coordinates and t-statistic of the voxel with the highest test result within each region. This approach is attractive because it provides an intuitive and fairly complete overview of the results of a study. Unfortunately, scientifically speaking, this approach also has some important disadvantages, that are sometimes overlooked. The most important disadvantage is that the resolution at which the results are published is too high to provide quantitative information about all the tested data. The most important consequence is that it is not possible to objectively determine if an fMRI study replicates or falsifies previous studies. Another consequence is that this method limits the possibilities to quantitatively compare signal changes between different regions in the brain. Last, but not least, this approach limits the possibility to quantitatively combine results from different studies to build more complex theories.

Methods:

We propose to introduce a system to quantitatively communicate group averaged fMRI results, by adding an extra layer to the MNI atlas, specifically designed for fMRI results. Starting at the MNI origin coordinate, the MNI brain is divided in isotropic regions of 20*20*20 mm (see figure 1). If needed, a region can be subdivided in 8 sub-regions of 10*10*10 mm, decoded A-H. This fMRI-MNI coordinate system can be used to communicate and replicate exact regions where a signal change was measured. Thus, it enables quantitative replication or falsification of previous fMRI studies, the combining quantitative results of several studies to build and test more complex cognitive models. We tested the potential value of this system on a dataset gathered in seven MRI centers in Europe. Scanners differed in brand (Philips, Siemens, GE) and field strength (1.5 – 3.0 T). Scans differed in repetition time (0.75 – 2.5 sec) as well as pulse sequence (PRESTO, EPI). Each center conducted a verb generation (VG) study in six subjects (M/V: 3/3). For each center, we compared the signal change in the left and right hemisphere in identical ROIs based on the proposed fMRI coordinate system ROIs were relevant for activity associated with Broca and its right hemisphere homologue (see figure 1, grey square, |t| > 5; N = 42).

Results:

In all 7 centers, we observed a significantly higher signal change in the left hemisphere compared to the right (p < 0.01; left hemisphere: 0.05%, right hemisphere: 0.52%, difference: 0.47%, sd: 0.07). Signal change differed more over centers than signal changes in the language task (motor: 1.15%, sd: 0.44; language:).
Supporting Image: MNI20_vg.png
   ·Figure 1: fMRI coordinate system overlay on verb generation activation pattern (N = 42)verb gemneration
Supporting Image: figs.png
   ·Figure 2: signal change (SEM) in Broca and homologue right hemisphere for each site
 

Conclusions:

We propose a new system to communicate group-averaged fMRI results that allows for quantitative comparison and verification of fMRI studies. Our multi-site study results confirm the potential value of this system. Signal changes measured with fMRI in identical regions appear to be highly reproducible over centers, even for small sample sizes. Our proposed system provides the possibility to quantitative compare, replicate, and falsify fMRI results, and can thus considerably aid in advancing the scientific quality of fMRI.

Higher Cognitive Functions:

Higher Cognitive Functions Other

Imaging Methods:

BOLD fMRI 1

Informatics:

Brain Atlases 2
Databasing and Data Sharing

Language:

Speech Production

Keywords:

FUNCTIONAL MRI
Language
Statistical Methods

1|2Indicates the priority used for review

Would you accept an oral presentation if your abstract is selected for an oral session?

Yes

I would be willing to discuss my abstract with members of the press should my abstract be marked newsworthy:

Yes

Please indicate below if your study was a "resting state" or "task-activation” study.

Task-activation

By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute the presentation in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels or other electronic media and on the OHBM website.

I accept

Healthy subjects only or patients (note that patient studies may also involve healthy subjects):

Healthy subjects

Internal Review Board (IRB) or Animal Use and Care Committee (AUCC) Approval. Please indicate approval below. Please note: Failure to have IRB or AUCC approval, if applicable will lead to automatic rejection of abstract.

Yes, I have IRB or AUCC approval

Please indicate which methods were used in your research:

Functional MRI

For human MRI, what field strength scanner do you use?

If Other, please list  -   multisite: 1.5 - 3.0T

Which processing packages did you use for your study?

FSL
Other, Please list  -   IDL

Provide references in author date format

No references