[FieldTrip] Help with ft_sourcestatistics

Fri Nov 18 15:09:28 CET 2022

Hi Francisco,

The fieldname ‘avg’, in the source reconstructed data has a historical reason, but is a bit unlucky and misleading. Indeed it refers to the fact that the numeric data in the sub-structure reflects an average (across a bunch observations). In your case it is the signal power in the baseline and in the post stimulus respectively.

Now, if you want to run a permutation test, you need a sufficient amount of subjects (which in this case serve as the ‘observations’). This is what Stephan refers to in his earlier reply in this thread. Given that you used only 1 subject worth of data, the code throws an error (and correctly so), because there is not much to permute. To do a meaningful statistical inference you need at least results from 15 or so (but ideally more) subjects.

If, on the other hand, you want to perform the statistics within the subject (i.e. using the individual trials as units-of-observation for the permutation) you need to set up your analysis in a slightly different way.

Best wishes,

Jan-Mathijs

On 18 Nov 2022, at 14:20, FRANCISCO JAVIER GÓMEZ CAMPOS via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:

Dear Stephan,

Thank you for your response, but in this case our source analysis data does not contain the average across subjects. If you refer to the ‘avg’ structure, it is one of the outputs coming from performing the source analysis by contrasting the pre and post stimulus interval. During the process, I calculate an inverse filter calculated from both intervals together and then applied separately to each interval, from this process I obtain the output ‘avg’, I do this for each subject and each condition.

You can see it in more detail at this link https://www.fieldtriptoolbox.org/tutorial/beamformer/<https://urldefense.com/v3/__https://www.fieldtriptoolbox.org/tutorial/beamformer/__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1IE_2mg0g$> in the section ‘Source Analysis: Contrasting activity to another interval’.

Our source data are distributed as follows, for each subject I have the mean power of each voxel for each condition independently, perhaps best understood with the following image: <https://urldefense.com/v3/__https://gyazo.com/00af9e5ca9647bec71d561c4197eefc5__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1KfbWELHw$> https://gyazo.com/39bb9c919f469d95f984a68b6dca6e18<https://urldefense.com/v3/__https://gyazo.com/39bb9c919f469d95f984a68b6dca6e18__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1JEfrkLwg$>

[https://lh6.googleusercontent.com/fLsaCgJYAhYIaQRGNxzNEdwzQN7n10QBbjS-oOeTbhvP68qQ3rBW8dnFa37-Vg2pzaNbY2NiVPCW4YCNLwwed-IpPQfXo-RzL1Sh9ac7LgqEkClX4kIr79C_BAfOvGhmtIMeW-wgY5VYJ31TSbyQpcllTBTnD13S5roabFDKvjtghYDMV7sr-2Kz1Hanbg]
We suspect that the problem could stem from the fact that our data structure differs from the one described in the tutorial for ft_sourcestatistics in that we do not provide sources from each trail, but the average source of each condition instead. So, our question is whether with this data structure we can perform a nonparametric randomization test, by creating the null hypothesis distribution shuffling voxels between conditions and then obtain an 'average' null distribution across subjects. I am not sure if this makes sense. Another way could be to shuffle conditions across subjects and obtain a null hypothesis distribution from this shuffle.

Thank you again,

Best,

Francisco Javier Gómez

Instituto de Biomedicina de Sevilla(IBIS)

________________________________
De: STEPHAN MORATTI <smoratti at psi.ucm.es<mailto:smoratti at psi.ucm.es>>
Enviado: jueves, 17 de noviembre de 2022 14:44
Para: FieldTrip discussion list <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>>
Cc: FRANCISCO JAVIER GÓMEZ CAMPOS <fjgomez-ibis at us.es<mailto:fjgomez-ibis at us.es>>
Asunto: Re: [FieldTrip] Help with ft_sourcestatistics

Dear Francisco,

It seems that in your source data you probably have the average across subjects, but the individual source data for each subject is missing. You may store them in individual cells and then submit these cells to the source statistics.

Best,

Stephan

El jue, 17 nov 2022 a las 11:32, FRANCISCO JAVIER GÓMEZ CAMPOS via fieldtrip (<fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>>) escribió:
Dear community,

I am having problems using the ft_sourcestatistics function. I would like to use this function
to conduct a within-subject analysis to establish which areas show differences between two
experimental conditions. I mainly have doubts about two aspects, the first one is about how to
attach the source analysis data, and the second one is about the design of the design matrix.

Regarding the data coming from the source analysis, I have the sources separated by
conditions (obtained by following this tutorial https://www.fieldtriptoolbox.org/tutorial/beamformer/<https://urldefense.com/v3/__https://www.fieldtriptoolbox.org/tutorial/beamformer/__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1IE_2mg0g$>):

Source analysis data for each condition:
freq: 4.8302
cfg: [1×1 struct]
dim: [85 110 132]
inside: [1234200×1 logical]
pos: [1234200×3 double]
method: 'average'
avg: [1×1 struct]

avg:
pow: [1234200×1 double]
noise: [1234200×1 double]
filter: {1234200×1 cell}
label: {61×1 cell}
filterdimord: '{pos}_ori_chan'

The code I use is the following (based on this code in this tutorial
https://www.fieldtriptoolbox.org/example/source_statistics/<https://urldefense.com/v3/__https://www.fieldtriptoolbox.org/example/source_statistics/__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1IM6xnJLg$> ) :

design=[1 2]; %for 2 conditions
% run sourcestatistics using cluster based correction %
cfg = [];
cfg.dim = sourceanalysis_condition1(1).dim;

cfg.method = 'montecarlo';
cfg.statistic = 'ft_statfun_depsamplesT';
cfg.parameter = 'pow';
cfg.correctm = 'cluster';
cfg.numrandomization = 'all';
cfg.alpha = 0.05; % note that this only implies single-sided testing
cfg.tail = 0;
cfg.design(1,:) = [1:length(find(design==1)) 1:length(find(design==2))];
cfg.design(2,:) = design;
cfg.uvar = 1; % row of design matrix that contains unit variable (in this case: trials)
cfg.ivar = 2; % row of design matrix that contains independent variable (the conditions)
stat = ft_sourcestatistics(cfg, sourceanalysis_condition1, sourceanalysis_condition2)

When I run this code, I get the following error:
Error using ft_statfun_depsamplesT (line 83)
The data must contain at least two units of observation (trials or subjects).

I understand that my design matrix and/or the structure of sourceanalysis_condition are
wrong, I have tried different combinations and I cannot find the solution. I hope you can
guide me in this regard so that I can adapt it to the conditions of my study.

Thanks,
Francisco Javier Gómez,
Instituto de Biomedicina de Sevilla (IBIS)
_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202<https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1IhCnYeZw$>

--

Stephan Moratti, PhD
Profesor de Psicología Experimental
Departamento de Psicología Experimental
Facultad de Psicología
Center for Cognitive and Computational Neuroscience
Universidad Complutense de Madrid
smoratti at ucm.es<mailto:smoratti at ucm.es>

[http://www.ucm.es/logo/ucm.png]

La información contenida en este correo es CONFIDENCIAL, de uso exclusivo del destinatario/a arriba mencionado. Si ha recibido este mensaje por error, notifíquelo inmediatamente por esta misma vía y proceda a su eliminación, ya que ud. tiene totalmente prohibida cualquier utilización del mismo, en virtud de la legislación vigente.

Los datos personales recogidos serán incorporados en la actividad de tratamiento 'Correoweb', bajo la titularidad del Vicerrectorado de Tecnologías de la Información, y en él el interesado/a podrá ejercer los derechos de acceso, rectificación, supresión u oposición ante el mismo (artículos 15-21 del Reglamento UE 2016/679 de 27 de abril de 2016).

[http://www.ucm.es/logo/ambiente.png] Antes de imprimir este correo piense si es necesario: el medio ambiente es cosa de todos.

This message is private and confidential and it is intended exclusively for the addressee. If you receive this message by mistake, you should not disseminate, distribute or copy this e-mail. Please inform the sender and delete the message and attachments from your system, as it is completely forbidden for you to use this information, according to the current legislation. No confidentiality nor any privilege regarding the information is waived or lost by any mistransmission or malfunction.

The personal data herein will be collected in the file "Correoweb", under the ownership of the Vice-Rectorate for Information Technologies, in which those interested may exercise their right to access, rectify, erasure or right to object the contents (article 15-21 of Regulation (EU) 2016/679, General Data Protection Regulation).

[http://www.ucm.es/logo/ambiente.png] Before printing this mail please consider whether it is really necessary: the environment is a concern for us all.

_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!6N5sAUOLFHX5VmnV9FuI390JcfBKDqKW5I94Eju8n6bs515XFjjjZfbDgj9-ZJOis1KRANuRDSZRpVED-QyvuI2uvBEic1IhCnYeZw$

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20221118/c672a283/attachment.htm>