Display results for permutation test

gnicolas1223 · May 24, 2022, 10:50am

Hello,

I am doing independent permutation tests on my sources scout and on my 29 subjects.Each subject has a individual cortex.
The permutation test goes well but the results come out as a matrix and not as a source map has explained in your tutorial.

So I cannot represented the results of the permutation tests on the cortex. I think it is because each subject has is individual cortex, I am right ?

Another thing, when I display the results as time series, the y-axis is labelled as 'No units', so I wonder what is the y axis ?

Thank you in advance for your response,
Gaëlle

Francois · May 24, 2022, 12:35pm

The permutation test goes well but the results come out as a matrix and not as a source map has explained in your tutorial.

If you select source maps in input of the test, you get source maps in output:
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics#Example_4:_Parametric_test_on_sources

If you select scouts time series in input, you get scouts time series in output:
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics#Example_5:_Parametric_test_on_scouts

If you want to display these scouts results on the cortex, the tutorial has a dedicated section for that:
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics#Convert_statistic_results_to_regular_files

the y-axis is labelled as 'No units', so I wonder what is the y axis ?

This is the T-statistic.

gnicolas1223 · June 16, 2022, 1:19pm

Thank you for your response.I managed to do my permutation tests on sources between two conditions for a given scout. Now I want to do permutation tests on sources between two scouts for a given condition. Is it possible to do that ?

Thank you in advance for your response,
Gaelle

Francois · June 16, 2022, 1:40pm

No, it is not possible do to this with the interface.

Comparing the min norm values between two different regions is not necessarily meaningful: testing that one region has higher values than another would mostly inform you about properties of the imaging technique that is used, not about the activity of the brain.

gnicolas1223 · June 17, 2022, 7:46am

Thank you for your response.

I try to script the permutation test with the function bst_permtest.mat but I having a problem.

[pv,S0,nGoodA,nGoodB,PS] = bst_permtest(A, B, TestType, dimPerm, nPerm, tails, isZeroBad)

I have two matrix A and B of 26 subjects x 1200 times samples (one for each ROI) as input matrix. I don't understand what number I must put for the parameter dimPerm ?

I try to put 1, 2, 3 but each time I have an error.

I try through the interface by putting breakpoints but it didn't help me solve my problem.

Thank you for your help,
Gaelle

Francois · June 18, 2022, 1:03pm

I try to script the permutation test with the function bst_permtest.mat but I having a problem.

I recommend you use only the high-level process call.
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting

I don't understand what number I must put for the parameter dimPerm ?

In this variable, you must report along which dimension you concatenated the various repetitions.

github.com

brainstorm-tools/brainstorm3/blob/master/toolbox/process/functions/process_test_permutation2.m#L165


      
              else

                  FreqList = TfMat.Freqs;

              end

              if (abs(OPTIONS.FreqRange(1) - FreqList(1)) < 1e-4) && (abs(OPTIONS.FreqRange(2) - FreqList(end)) < 1e-4)

                  OPTIONS.FreqRange = [];

              end

          end

          

          % ===== INPUT DATA =====

          % Concatenante values in the 4th dimension

          OPTIONS.Dim = 4;

          % Set all the bad values to 0

          OPTIONS.isBadZero = 1;

          % Call extraction process: FilesA

          [sDataA, OutputType, matName] = process_extract_values('Extract', sProcess, sInputsA, OPTIONS);

          if isempty(sDataA)

              bst_report('Error', sProcess, [], 'No data read from FilesA.');

              return;

          end

          % Read FilesB

          sDataB = process_extract_values('Extract', sProcess, sInputsB, OPTIONS);

I try to put 1, 2, 3 but each time I have an error.

I guess the previous comment should help you solve your problem.
For future bug reports: please include the full error messages you get when you report an error.