 |
 |
You can pretty much do whatever you want, as long as you have a hypothesis and it is testable. And as long as you don't, say, look at the results and pick the location that has the highest t-values. Chris ________________________________________ From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of Andy Yeung [[log in to unmask]] Sent: Thursday, January 09, 2014 10:13 PM To: [log in to unmask] Subject: Re: [SPM] Small Volume Correction, and Video Tutorial wanted Dear all, Thanks very much for all of your advice. I have searched the literature for respective ROIs/MNI coordinates. Is it acceptable if I just pick one ROI/VOI out of several reported from past meta-analysis? Or do I have to apply all reported ROI/VOIs into my own SVC/ROI? Thanks a lot. Best, Andy On Fri, Jan 10, 2014 at 7:41 AM, Andy Yeung <[log in to unmask]<mailto:[log in to unmask]>> wrote: Dear Dr Goto, Below is forwarded messages from SPM mailing list talking about small volume correction. They say we have to define VOI before looking at the actual results, just like what we have discussed about. This is by either: 1. anatomy 2. previous data 3. previous coordinates reported by other papers We can make a mask consisting multiple VOIs. We may perform multiple small volume correction for each contrast, but if so, we have to lower p value threshold accordingly. Best, Andy ---------- Forwarded message ---------- From: MCLAREN, Donald <[log in to unmask]<mailto:[log in to unmask]>> Date: Thu, Jan 9, 2014 at 10:49 PM Subject: Re: [SPM] Small Volume Correction, and Video Tutorial wanted To: [log in to unmask]<mailto:[log in to unmask]> One additional point to add to what Chris said, you should only perform SVC once per contrast. If you want to look at multiple areas, then you either need to: (1) Form a single small-volume mask that includes all the regions of interest; or (2) Divide the SVC threshold by the number of times you've performed the SVC. Otherwise, the significance of what you find will be inflated if you repeatedly perform SVC on your data. 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<tel:%28773%29%20406-2464> ===================== On Thu, Jan 9, 2014 at 7:11 AM, Christophe Phillips <[log in to unmask]<mailto:[log in to unmask]>> wrote: Dear Andy, you should NOT define your SVC's based on the uncorrected results. This means, in your case, you should not simply pick a peak voxel and define a spherical volume around it for small volume FWE. The a priori volume/location should be defined before looking at the results and independently from these, for example based on - some anatomical priors, - coordinates reported in the literature - your own results from a different experience. The size of the SVC should also be defined before hand and be representative of the prior believe about the activation location. HTH, Chris Le 9/01/2014 11:57, Andy Yeung a écrit : Dear all, I've read SPM manual which says: 31.3.7 Small volume correction If one has an a priori anatomical hypothesis, eg. in the present example Broca's area will likely be activated during word generation, one may use the small volume correction option. Press the \small volume" button in SPM Interactive (bottom left) window and select a suitable region, e.g., a 30mm sphere with its centre at 44 16 0. The region can also be de ned using mask images derived from previous imaging data. The corrected p-values will change, as shown in Figure 31.11. For example if I got 4 peak voxels at 2nd level analysis, could I perform SVC 4 times separately at those peak voxels in order to report them with smaller FWE p values? If yes, how big should I set the diameter of the SVC? Thanks very much for your time! Yours, Andy