Tuesday, November 20, 2012

FSL Tutorial: Featquery_gui

Now that we've created our masks, we can go ahead and extract data using FSL's featquery tool. You may want to run it from the command line when batching large numbers of subjects, but this tutorial will focus on Featquery_gui, a graphical interface for loading subjects and ROIs, and then performing data extraction from that ROI. The procedure is similar to Marsbar, and I hope that the video is clear on how to do this.

Also, I've attached a Black Dynamite video for your enjoyment. Nothing to do with ROIs, really, but we all need a break now and then.




10 comments:

  1. This post is really helpful, thank you. Do you typically take the %signal change values from each individual if carrying out additional analyses? I've come across some info from Mumford (2009) which mentions that the scaling used across individuals is likely to be different and so featquery is not recommended to be used to compare across individuals. Do you have any thoughts on this? Many thanks.

    ReplyDelete
    Replies
    1. Hi there,

      Thanks for the heads-up; I have usually used featquery to extract data from each subject at the second level, but the percent signal change document on her website gives some good reasons for her alternative method using fslmeants.

      I'll probably write about that in a future post, or perhaps get Dr. Mumford to make a guest post about it.


      Thanks!

      -Andy

      Delete
    2. Hey!

      I know this is a really old post so don't expect a reply, but just wondered your opinion on the following analysis.

      I'm aiming to show that amygdala shows more activation to violent pictures (compared to non-violent pictures) in males compared to females.

      I am not entirely sure how to conduct an ROI analysis here though.

      Would I conduct first-level analysis across all participants using the violent > non-violent contrast, then perform group analysis to compare males with females? If so, where would the ROI come in?

      Any help would be appreciated.

      Delete
    3. Hey there,

      Your intuition is on the right track; you should do a first-level contrast for each subject, and then select each subject's 1st-level FEAT directory for the ROI analysis.

      To do an ROI analysis, you don't necessarily need to do a second-level analysis on the neuroimaging data. Second-level analyses are useful for presenting whole-brain analyses of contrasts, as well as creating group-level functional localizers. But for most ROI analyses, you just need the 1st-level contrasts, and then a mask to extract those from.

      -Andy

      Delete
  2. Great to hear what others are doing. Many thanks

    ReplyDelete
  3. Great blog Andy! I have a quick query....how do you interpret the % signal change values output? Are the values scaled between 0 and 1 so that a change of say 0.1 can be considered a 10% change? Or are the values scaled between 0 and 100?

    Thanks!

    ReplyDelete
    Replies
    1. Hey Ticker,

      It should be on a scale from 1 to 100. So if it gives you a value, say, of 5, then that is a 5% signal change.

      Personally, I think it is more useful to just have a contrast or beta estimate, without the conversion to percent signal change. You can change off this option in the FEAT GUI.

      -Andy

      Delete
  4. Hi Andy! Your Blog is awesome! Thank you so much for helping us new beginners out with your videos and posts.
    I still try to figure out a lot of things in FSL and have a very important question. I don't know if this actually fits here, but I hope you can help me out here.
    I am analysing my data with FEAT and apparently, there is 1 volume missing in one participants data set. Do you know what I can do to still use the data? Is there a way to "create" a volume, if that makes any sense? ;)
    I would reeeally appreciate your help.

    Kind regards,
    B.

    ReplyDelete
    Replies
    1. Hey B.,

      Is FSL throwing an error when you try to run it with one volume missing? I don't think it should be a problem, but if it really is intractable, you may consider dropping the volume in the other runs (I'm assuming it's at the end or something, where it can't cause too much mischief).

      -Andy

      Delete
  5. Hi Andrew,

    I'm working with an inherited data set wherein the first two levels of analysis have already been done, but the previous contributor did not generate cope files for each contrast of interest. Instead, each participant's cope file specifies the average activation for each of 6 EVs of interest. Do you think I can still use these files for ROI analysis in Featquery? I only have a few weeks with this data and I don't have time to redo the level-one analysis from scratch. Thanks very much, your blog has already been a big help this summer!

    -T

    ReplyDelete