 |
 |
Dear all. I tried everything you told me and failed. When I insert scans using select all it tells me that I should define factors levels using a nscans times 4. What's this matrix supposed to be? Why 4? I have only two factors! Inviato da Outlook Mobile<https://aka.ms/blhgte> On Mon, Feb 1, 2016 at 1:48 PM -0800, "PRESOTTO LUCA" <[log in to unmask]<mailto:[log in to unmask]>> wrote: Thanks a lot! Don't know how the "specify all" selection slipped :( I swear I was looking for it. I'll look into the full factorial design then. I just have a concern with it. One of the factors does not different levels of the same stimulus but many (more than 2) different groups of subjects (e.g: G1: Alzheimer's patients, G2: healthy controls, G3: DLB patients etc...etc...). Is letting SPM compute "interactions" correct in this case? ________________________________ Da: Angstadt, Mike <[log in to unmask]> Inviato: lunedě 1 febbraio 2016 22.38 A: PRESOTTO LUCA; [log in to unmask] Oggetto: RE: [SPM] Flexible factorial, full factorial or GLM flex? You should use the “Specify All” option rather than “Specify Subjects.” This will allow you to select all of your images at once, and enter the full condition matrix at one time rather than subject by subject. I believe SPM treats “subject” as a special condition name, which can impact where the columns are in your eventual design matrix. For the non within-subject design that you mention at the end, you’d be better off just using the Full Factorial design specification, not the flexible factorial. The Full Factorial will have you specify factors, and then enter images by cell and will automatically create the main effect and interactions contrasts for you. -Mike -- Mike Angstadt Research Computer Specialist / PANLab Lab Manager Department of Psychiatry / University of Michigan (734) 936-8229 From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of PRESOTTO LUCA Sent: Monday, February 01, 2016 4:05 PM To: [log in to unmask] Subject: Re: [SPM] Flexible factorial, full factorial or GLM flex? Hi Donald Thank you a lot for your answer. Can I ask you to clarify a bit how to set everything up? I've done a lot of trials and errors attempts and things don't add up that much. The first question is how to insert scans. Let's suppose I've got 2 factors + the subject factor. If this is the case I should declare 3 factors and then do "new subject" in the batch for each subject (really? That's going to take half on hour for an experiment with 50 subjects). Then in the conditions field if I add a 2xn Scans matrix, since SPM is indexing automatically the patient. Is this correct? I can't just do a single "new subject" and put all hundreds of scans there and then feed a matrix including also the patient index, can I? I tried and it just discards the last factor of my "conditions" matrix. Then, when I set up the "main effects & interactions" if my first factor was "subject", than G, than C and I setup the main effects as "1", than "2", than "3", I get in the design matrix the subject factor as the last rows, not the first ones. The opposite of you if I understood it right. Also, I don't get how do you get so many columns in the design matrix. Shouldn't they just be N (subjects)+ sum(factor_levels)? Therefore 27+2+2? BTW, on top of understanding how it works in general, currently I'm dealing with a design where "subject" is not a factor (i.e.: each groupXcondition set of scans is made of different subjects). How should I set it up? Click "new subject" just once and dump all the scans there? And how should I index them if the last factor index gets dropped? Thank you again! Luca ________________________________ Da: MCLAREN, Donald <[log in to unmask]<mailto:[log in to unmask]>> Inviato: lunedě 1 febbraio 2016 19.21 A: PRESOTTO LUCA Cc: SPM Oggetto: Re: [SPM] Flexible factorial, full factorial or GLM flex? Luca, I've always used the Specify All option to list my design matrix and scans. For each scan, you need to have a row in the factor matrix. Column 1 should be observation number, column 2-4 should correspond to factors 1-3. Each value is the level of each factor for the corresponding scan. Put all main effects into the model, including the subject factor. For the contrasts: Here is an example for the interaction of a 2x2 ANOVA. This is for a design with 18 subjects in group 1, 9 subjects in group 2, 2 group terms and 2 conditions: Start with the simpliest element, single subject in a single condition, build its contrast, repeat for all subjects and conditions, and then combine the ones you want. S1G1C1=[1 zeros(1,26) 1 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0] S1G1C2=[1 zeros(1,26) 1 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0] .... Now average your G1C1 and by summing and dividing by the number of subjects, you'd get G1C1=[ones(1,18)/18 zeros(1,9) 1 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0] and G1C2=[ones(1,18)/18 zeros(1,9) 1 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0] and G2C1=[zeros(1,18) ones(1,9)/9 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0] and G2C2=[zeros(1,18) ones(1,9)/9 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0] Now subtract G1C1-G1C2 AND G2C2-G2C1 G1C1-G1C2=[zeros(1,27) 0 0 1 -1 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 0 0 0 0] and G2C1-G2C2=[zeros(1,27) 0 0 1 -1 0 0 0 0 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0] Now subtract these two: Interaction contrast=[zeros(1,27) 0 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 -1 1 0 0 0 0 0] Best Regards, Donald McLaren, PhD On Sun, Jan 31, 2016 at 2:22 PM, PRESOTTO LUCA <[log in to unmask]<mailto:[log in to unmask]>> wrote: I strongly second this request. I also need to run a 3x2 flexible factorial design but I haven't found a single tutorial/document/manual about how to correctly insert scans (even if in the end I've figured it out) and, more importantly, how to write contrasts correctly. I've been working on this for a full week and it's driving me totally insane. Truly hope somebody can help! Cheers, Luca ________________________________ Da: SPM (Statistical Parametric Mapping) <[log in to unmask]<mailto:[log in to unmask]>> per conto di Deza Araujo, Yacila Isabela <[log in to unmask]<mailto:[log in to unmask]>> Inviato: domenica 31 gennaio 2016 00.08 A: [log in to unmask]<mailto:[log in to unmask]> Oggetto: [SPM] Flexible factorial, full factorial or GLM flex? Dear SPM experts, I am running a connectivity analysis with two groups and 3 pharmacological interventions in each group. The first level is already done, now my question is about the second level. Is it correct the use of a Flexible factorial design with 2 factors (2 and 3 levels respectively)? I expect to have “subject” as a factor, but according to what I´ve read this is not possible in this kind of designs. I also want to look at linear and quadratic effects. My alternative would be GLM Flex. I tried to run a model, but I really need support with the interpretation. Is there any manual or deep explanation about the results and their interpretation? I am looking for answers in the respective group, but I haven’t found anything clear and I thought that maybe someone with experience in this kind of design can help me or give me some advice about the results and the way to make the contrasts. Thanks a lot for your attention Yacila Deza Araujo Rispetta l’ambiente: non stampare questa mail se non č necessario. Respect the environment: if it's not necessary, don't print this mail. Rispetta l’ambiente: non stampare questa mail se non č necessario. Respect the environment: if it's not necessary, don't print this mail. ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues