Ft_sourcestatistics

Dear Francois and all,
I am sorry for so many questions.
Recently, I have been learning how to conduct PAC value and do statistical analysis at source level of epileptic though Brainstorm. My operation procedure is as follows:

  1. Import the MRI of each individual
  2. Import .fif MEG file of each subject and refine MRI registration using head points
  3. Add event of spike
  4. Preprocess data using 0.1-70Hz band-pass filter, 50Hz notch filter and ICA
  5. Import in dataset using spike event
  6. Generate cortex surface using Overlapping spheress
  7. Select "No noise modeling" to prepare noise covariance
  8. Compute soure using the following settings
    图片17
  9. Extract source estimation to Matlab and compute the PAC of theta-gamma using modulation index
  10. Replace the field “ImageGridAmp”, “Comment”, “Time” with PAC value, “'Extract values: [-0.000s,0.000s]” and 0 respectively
  11. Import modified structure from Matlab
  12. Project source to default anatomy
  13. Put the subjects into their respective groups in process2 and select ft_sourcestatistics
    Because I just want to learn the process and want to see the results, I set the p-value and cluster alpah as 1. The results are as follows.
    图片18
    However, when I open the “FT t-test unequal cluster [0ms]: sub01/Extract values: [-0.000s,0.000s] (2) vs. sub02/Extract values: [-0.000s,0.000s] (5)”, it gives the message as follow:
    BST> All values are null. Please check your input file.
    The Clusters on the right panel are also empty. In addition, the Time is displayed as [0,1]ms
    图片19
    Sorry this is so long. I unable to figure out the problem or if I am doing anything wrong. Could you check it for me and point out the mistakes. Thank you in advance for your assistance.
    Best,
    Liu Xinyan

Checked how your band-pass filter alters the shape of your spikes before using this filter blindly.
Make sure the ICA components you are removing do not include any of the spiking activity you are interested in.

BST> All values are null. Please check your input file.

This means nothing is significant in your test.
I'm sorry I can't help you much with FieldTrip functions. If you have specific questions about the inputs of ft_sourcestatistics, please ask them on the FieldTrip support channels (put a breakpoint where this function is called in process_ft_sourcestatistics.m in order to report all the values used in input):
https://www.fieldtriptoolbox.org/support/

In addition, the Time is displayed as [0,1]ms

This means there is no time dimension in your dataset.
This seems correct for a PAC analysis...

Dear Francois,
Thanks for your advice. I have one more question though.
I set cluster alpah as 0.05 and number of randomizations as 1000 when I use ft_sourcestatistics. The command window shows that 16 positive clusters and 377 negative clusters were found, as shown in the diagram below.
图片20
However, when I set the significance level α as 1 in the Stat tab, only 6 clusters are shown on list of Clusters.
图片21
So, what kind of clusters will be displayed on list of Clusters? Is there any other factor besides the level of significance?
Thank you again for all the help.
Best,
Liu Xinyan

Even when you set "significant level alpha" to 1, Brainstorm selects only the clusters with p<0.999, in order to avoid displaying these 400 useless clusters. See the selection code:

I get it now. Thank you, Francois :coffee: :coffee:.