 |
 |
Hi, Glm is a GUI itself so I'm not sure if you just mean you started it from the main fsl GUI or whether you were trying to get a non-GUI version initially. There is only a GUI version. When you run it (Glm_gui on a mac or Glm on other machines) it will bring up two windows and you need to go to the Higher-level mode and fill in both the EVs and Contrasts tab. Once you've done this and viewed your design, saving it should create a set of files, all starting with the name you gave. So if you called it "mydesign" then it should save (among others) mydesign.fsf, mydesign.mat and mydesign.con. You will need the latter two for randomise. In randomise you specify the mydesign.mat file for the -d option and the mydesign.con file for the -t option. See the example in the web page documentation. It seems that you have stopped short of entering the whole command in the example you included here. As for running the stages, you should have all the images ready and run each stage once but include the set of images you are working with. Again, see the web page docs which explain this, or even the fslcourse practical which takes you through some of the steps on our example data. All the best, Mark On 28 May 2009, at 07:09, Nick Papanikolaou wrote: > Dear All, > This is my first experience with TBSS analysis. I followed step by > step the instructions > found on the web site with success until the last step where I have > to generate the > design.mat and design.con files. I tried on the terminal to run the > Glm with no luck, so I > used the GUI, I defined my contrasts and everything but when I > pressed save I could see > only a design.fsf file, furthermore by trying the following: > randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d > design.mat -t > > I got: > -t: Missing non-optional argument! > try: randomise --help > > Any help could be appreciated. > > P.S. I have some questions about the workflow of TBSS. For example > if I want to > compare one patient each time with a group of normal controls, do I > have to repeat the > steps from tbss_1_preproc ... to tbss_4_prestats for all the > subjects since the controls will > be the same in multiple experiments. >