Dear Joaquim,

I'm starting my meta-analysis with SDM, which looks really good! Thanks for creating this tool :)

I have read in this forum (2015) that you would like to implement the analysis of several contrasts from one paper in SDM. I would like to know if it's already possible or if we need to do some separate .txt per each contast, like if they were different papers?

Thank you in advance,

Fins aviat,

Mónica Franco Emch

I'm also interested in this question, as including each contrast as if they were separate studies would mean repeating the participant cohorts, and I'm unsure about this?

Many thanks,

Aisling

I also encountered the same situation。。it can be defined as repeated measures?

I read this two Repeated measures papers:
Alegria AA, Radua J, Rubia K. Meta-analysis of fMRI studies of disruptive behavior disorders. Am J Psychiatry 2016;
Norman LJ, Carlisi C, Lukito S, Hart H, Mataix-Cols D, Radua J, Rubia K. Structural and functional brain abnormalities in attention-deficit/hyperactivity disorder and obsessive-compulsive disorder: A comparative meta-analysis. JAMA Psychiatry 2016; 73:815-25. https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2526239
 In supplement, I see the sentence 'This new simple approach will be included in the next version of SDM software to allow other researchers conduct repeated-measures meta-analyses.'
I'm not sure, the 5.141 is the 'next vesion'?

v5.11 - Major changes version. Main improvements: New improved graphical
(08/16) user interface for Windows, Linux and Mac OSX, possibility to
combine repeated measures (e.g. from several contrasts of the
same sample), easy automatic creation of funnel plots and Egger
tests. Other improvements: recreated t-maps are also saved (along with
the effect size maps), threshold procedures create a mask for each peak
(rather than for each blob), added "other_brett" for FreeSurfer. Solved
a small bug in the automatic creation of masks when thresholding SDM
results (masks were sometimes empty).

There's an option to combine images in the new version of SDM, but not sure if this applies to peak coordinates .txt files, as well as statistical maps.

Dear Mónica, Aisling, Hui,

as you stated, you must combine the images into a single one. Thus:

1. Recreate the t-maps of the contrasts of the study with two contrasts. To this end, you can conduct the pre-processing step of a "meta-analysis" of these contrasts.
2. Combine the t-maps generated in this "meta-analysis", obtaining a single combined t-map for the study.
3. Conduct the meta-analysis of all studies, including only the single combined t-map for the study with two contrasts (along with the automatically created "var_reduction" text file).

Hope this helps,

Joaquim

Dear Joaquim

If there are 3 contrasts in one paper , the third of which has no significant difference, should I merge 3 T-maps or just 1 + 2？

I try to combine 2 T-map which recreate by ES-SDM5.1.41。 But I didn't success. I don't know if there is a problem with my steps。

• I make sdm table and corresponding txt by each contrasts， do the processing。
• combing images，for example Map 0，t _ChunA2015.nii.gz; Map 1, t_ChunB2015.nii.gz
• I click OK, the software just stop in Imgcalc exection 0/2 0%; Variance reduction creation 0%

Dear Hui,

you should include all contrasts (as far as they are all within the scope of your meta-analysis), you should not exclude a contrast because it has no statistically significant findings.

Regarding the error, I'm sorry to hear that. Please tell us which operative system are you using and email us the data so that we can look at it.

With the best wishes,

Joaquim

Dear Joaquim

Thank you for your answer.
I find the WIN64 had been reported a bug in combine imges.https://www.nitrc.org/forum/forum.php?thread_id=8264&forum_id=3982
I replace the binaries of my package, It's success!

So If I want to Conduct the meta-analysis of all studies, including only the single combined t-map for the study with two contrasts.
I should

• Created A new folder.
• Convert the combined T-map.
• Put the converted T-map and t_Chun2015.var_reduction.txt with other txt in a new folde.
• Preprocessing again

How to use the t_Chun2015.var_reduction.txt?
I tried to change the file size from 0.65 to 0.8, which is determined by the assumed correlation and number of images, and found pp_Chen2015_var.nii.gz changed. I can think of combine images do not change the images just add them up?

Best,

Zheng Hui

Dear Joaquim,

but is it possible also to combine several contrasts for just coordinate-based meta-analysis or does it just work for image-based? If it's not possible with coordinate-based, can I at the end just control for it in a post-hoc analysis?

Thank you in advance,

Mónica

Dear Dr Hui,

SDM will use the "var_reduction" file automatically, you just have to put it in the main meta-analysis folder.

With the best wishes,

Joaquim

Dear Monica,

it is possible. You must first convert(*) the coordinates (and t-values) to t maps using the "pre-processing" button, and then combine these t maps to have a single t map. Finally, in the main meta-analysis you must only use this combined t map and an automatically generated "var_reduction" file.

Hope this helps,

Joaquim

* Please that the word "convert" here is not fully correct. The correct wording should be something such as: "You must first recreate the t-maps from the coordinates and t-values of the peaks of the clusters of statistical significance."

Dear Joaquim

The second step of the 'Convert images', because the use of original 'Combine images' (e.g, t_chun.nii.gz, 2contrasts combine) suggested that the log wrong.

Preprocessing studies
WARNING: Header of NIfTI file 'Dong2011A' specifies a statistic other than t. Please convert the image.
Preprocessing Monte Carlo permutation
WARNING: File 'sdm_maps.txt' not found.
Removing previous permutations
Calling web browser
Pre-processing finished!

Thank you for your patience,
Zheng Hui

Originally posted by Hui Zheng:

Dear Zheng Hui,

How did you generate the file Dong2011A? Can you describe the exact commands in sdm you run?

Alternatively, you could send us the source files and we'd try to find the error.

Best,

Anton

Originally posted by Hui Zheng:

Dear sdm-users and creators,

I hope you had nice holidays! Thank you for your involvment and pacience. After recreating the t-maps of the contrasts oft he study with several contrasts and combining the t-maps generated in order to obtain a single combined t-map, how can I include the information of each contrast in the meta-analyses table? I mean, I want to put the information of each contrast as a regressor in order to study if there are differences in the result. Or do I need to conduct separate sub meta-analyses each time with just a specific contrast? For example, if I want to study that in one contrast the stimuli was visual and in another the stimuli was verbal, how can I include this information in the table in the main meta-analyses in which I just have one t-map for visual and verbal stimuli type?

Thank you in advance,

Dear all again,

I also got the same as Zheng Hui, it says that I need to "convert image" after combining the studies that have more than one contrast. In the meta-analyses folder I have all the studies with one contrast (xxx.txt) and the .nii and .var_reduction from the combined studies. I am using Windows 64 and I already put the newest version of the software with the bin folder changed thanks to an answer from Anton Albajes-Eizagirre. Is someone not using Windows 64 having this problem? Or do I need to convert the image first in order to continue?

[u]log:[/u]

WARNING: Header of NIfTI file 'Walter2003_combination' specifies a statistic other than t. Please convert the image.
The program will close as this problem could cause fatal errors.

aPreprocessing Monte Carlo permutation 1 of 50
WARNING: File 'sdm_maps.txt' not found.
The program will close as this problem could cause fatal errors.

[u]aPreprocessing Monte Carlo permutation 2 of 50.....
[/u]

Dear Mónica,

I'm afraid that you can only use each study once, choosing whether to consider it either visual or verbal (or none, if you think original combined contrasts could collide).

You could run two metas. In the first one, using the original contrast for the visual stimuli and taking the study as visual (= marking it accordingly in the sdm table), and in the second using the original contrast for the verbal and taking it as a verbal study (again, in the sdm table).

I hope this helps, kind regards,

Anton



Originally posted by Mónica Franco Emch:

Dear Mónica,

Could you please send us the contents of the Execution Log (found in the menu bar Tools->View execution log) for the whole session? It looks something went wrong while converting the t-maps, but I'd like to check the whole process.

Kind regards,

Anton



Originally posted by Mónica Franco Emch:

Dear Anton,

I attached the file with the log result.

Thank you in advance,

Dear Anton,

thank you! I will then do a whole analyses with everything combined and then two separate ones in which in one I just include visual and in the other one just auditory stimuli contrasts.

Fins aviat,

Dear Mónica,

It seems the problem was in the combination step. Did you use the 'Convert images' command with the combined image (Crottaz-Herbette_combination)?

Hope this helps, kind regards,

Anton

Dear Anton,

I delated the Crottaz-Herbette_combination and put just one contrast instead of the two. And now, when I was doing it, Cairo_combination is the problem. I did then the preprocessing and then combination of the Cairo.txt files again, and put the Cairo_combination.nii and Cairo_combination.var_reduction back in the whole meta-analysis. I attached the log. It still not working. Is it possible that the problem is due to me using Windows 64, should I try sdm on other operating system?

Thank you for being so helpful :)

Dear Mónica,

After applying the comand 'combination', did you try to apply the command 'Convert image?'

Anton