 |
 |
On Fri, Jan 11, 2013 at 4:52 PM, Devin Sodums <[log in to unmask]> wrote: > Dear Dr. McLaren, > > Thank you very much for that, I feel quite silly that I was trying to do group analyses with just one subject. > > I was able to construct the large first level model with both scans in the model, as separate sessions. I was then going to do the same for the controls, and then was going to use the contrasts in a two sample t test. > > However, I was wondering if there was another route that was legitimate: > > For option (3) " -Use multiple first level models, subtract the 2 con images, then > use a two-sample t-test with equal variance to compare the results." > > My hope was to do the following: > > 1. Take con image from the first level model from scan 2 from a single control and subtract con image first level model from scan 1 from it. > 2. Do same for all controls (N=14, two scans) > 3. Then do same for individual patients (we have 5 total but each are so unique I cannot put them in one group) and take the subtracted con image from the patient and from the 14 controls and compare an individuel patient to the 14 controls using a two sample t-test with equal variance. > > Is this a legitimate approach? Yes. This is option 3. > > Thank you again for all of your help. > > Best, > Devin > > > ___________________________________ > > Devin Sodums > Research Assistant, Jones-Gotman Lab > Montreal Neurological Institute > 514 398 6644 ext. 00262 > > ________________________________________ > From: MCLAREN, Donald [[log in to unmask]] > Sent: Thursday, January 10, 2013 9:11 PM > To: Devin Sodums > Cc: [log in to unmask] > Subject: Re: [SPM] No significant voxels still after changing to non sphericity > > Devin, > > Thanks for the models, it was clear where the mistake occurred. > > You can not use group models with only 1 subject. I thought this might > be what you were doing, but you had also mentioned you had a group of > controls. > > If you want to compare two scans within a single subject, then you > need to build 1 large first-level model with both scans in the model, > but as separate sessions. Then form a contrast comparing the different > scans. > > If you want to compare the difference with that of controls, then you > have several options: > (1) Use larger first level model for all subjects and then use the > contrast and do a two-sample t-test with equal variance of the > difference images; > (2) Use multiple first level models and use a flexible factorial > design with 2 groups, groups should have equal variance. > (3) Use multiple first level models, subtract the 2 con images, then > use a two-sample t-test with equal variance to compare the results. > > All of these approaches should yield the same answer. > > Hope this helps. > > Best Regards, Donald McLaren > ================= > D.G. McLaren, Ph.D. > Research Fellow, Department of Neurology, Massachusetts General Hospital and > Harvard Medical School > Postdoctoral Research Fellow, GRECC, Bedford VA > Website: http://www.martinos.org/~mclaren > Office: (773) 406-2464 > ===================== > This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED > HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is > intended only for the use of the individual or entity named above. If the > reader of the e-mail is not the intended recipient or the employee or agent > responsible for delivering it to the intended recipient, you are hereby > notified that you are in possession of confidential and privileged > information. Any unauthorized use, disclosure, copying or the taking of any > action in reliance on the contents of this information is strictly > prohibited and may be unlawful. If you have received this e-mail > unintentionally, please immediately notify the sender via telephone at (773) > 406-2464 or email. > > > On Thu, Jan 10, 2013 at 1:39 PM, Devin Sodums > <[log in to unmask]> wrote: >> Dear SPM experts, >> >> I am trying to take a patient who has done the same test twice and run a flexible factorial design to see if there are any differences between scans. My end goal is to use the con.img's that would be outputted as another comparison with a group of controls that have also done the test twice. I have gotten the "no significant voxels." message each time I try comparing the single patient's scan 2 to scan 1. >> >> I have used a two-sample t-test, ANOVA-within a subject, and flexible factorial design, and each of them is still giving me the same error. After reading the posts about changing the independence and variance such that non-sphericitiy is assumed, yet I still receive the same message. I have checked the images with checkreg and they are the same dimensions and everything. I have also tried changing the defaults.stats.fmri.ufp to a more liberal threshold (from 0.001 to 0.01 and even to 0.1) but also still receive the same error message. >> >> This is the first time I have tried comparing a single subject's first and second scans, so perhaps I am doing something wrong to begin with? From what I have read, repeated measures analysis is appropriate, and I believe I set it up appropriately as well. >> >> Any advice/help would be greatly appreciated. >> >> Thank you, >> >> Devin