Back to PLS HelpPosted on 04/25/17 21:47:45
Number of posts: 3
Hi!
I have a problem of PLS (at least seemingly) randomly choosing either hot or cold colour to represent patterns of activation for 2 condition PET data, e.g. sometimes cold represents voxels of increasing intensity across conditions and sometimes vice versa. Is there a way to change the color mapping for significant voxels?
I'm running PLS on Matlab 2017b, OS X Sierra. Tried also with Matlab 2014a on Windows 7 with similar results.
Thank you!
/Oskari
Replies:
Posted on 04/25/17 22:05:00
Number of posts: 394
Hi!
I have a problem of PLS (at least seemingly) randomly choosing either hot or cold colour to represent patterns of activation for 2 condition PET data, e.g. sometimes cold represents voxels of increasing intensity across conditions and sometimes vice versa. Is there a way to change the color mapping for significant voxels?
I'm running PLS on Matlab 2017b, OS X Sierra. Tried also with Matlab 2014a on Windows 7 with similar results.
Thank you!
/Oskari
the math underlying PLS does not have insight into your design, so the patterns that emerge are the effects in your data. The interpretation is up to you. If you are facile in matlab, you can reverse the signs of the output by multiplying by -1 and then display the output. Alternatively you can export the results to another program to modify the colour scales.
Posted on 04/26/17 00:49:57
Number of posts: 3
Thank you for the prompt reply! I figured a simple -1 multiplication would do the job in Matlab but I don't know how to apply this. How can I find the output values which sould be reversed?
Posted on 04/26/17 06:46:15
Number of posts: 394
Thank you for the prompt reply! I figured a simple -1 multiplication would do the job in Matlab but I don't know how to apply this. How can I find the output values which sould be reversed?
take a look at the help file if you type: help pls_analysis from the matlab command line for the varaible list. The variable results.compare_u is the statistical (bootstrap) image. If you do the modify the results file, you should save it under a new name
I'm Online
Posted on 04/26/17 09:32:33
Number of posts: 229
and you have to multiply both sides of the equation by -1 (design saliences and bootstrap ratios [brain saliences])
Posted on 04/26/17 13:55:46
Number of posts: 3
Thank you!
Login to reply to this topic.
