Hi,
Please disregard my previous mail. I was a little too trigger happy..., the 1.928 threshold set by tksurfer is too high for any vertex to survive so the two methods DO concur. I should actually look at the maps and not just the output...
The reason I asked is that I'm combining images to obtain an overall FDR, either for both hemispheres or for several comparisons. I don't see how this can be done in tksurfer so I'm using Nichols' function, which works just fine. I just don't know how to get an FDR threshold when no vertex has a p-value that survives - but tksurfer obviously produces such a threshold. How do I obtain that threshold?
(Or is it simply set just higher than the highest -log(p-value) in the image, for display purposes? It seems always to be just above.)
thanks (and apologies for the confusion), LMR
From: larilin@gmail.com To: freesurfer@nmr.mgh.harvard.edu Subject: fdr thresholds in tksurfer compared to Nichol's function Date: Thu, 25 Jun 2009 14:07:08 +0200
Hi,
I've used the fdr function in tksurfer and compared the results to Tom Nichol's matlab function (http://www.sph.umich.edu/~nichols/FDR/). I use p = 10.^-abs(pfile), where "pfile" is 'sig.mgh'.
I find that the p-threshold (pID, assuming positive dependence) is usually exactly the same as the one in tksurfer (probably because you use the same algorithm..?).
But in some cases where pID is empty, which should mean that there is no suprathreshold vertex, I do get a threshold in tksurfer. Here's an example:
For one contrast where pID was returned empty in Nichol's matlab function, I got a threshold of 1.927741 in tksurfer.
Is this supposed to happen?
The output is:
MRISfdr2vwth(rate=0.05000, sign = 0, 1, only marked = 0) = 1.927741
surfer: MRISfdr2vwth with rate 0.05 and sign 0 returned threshold 1.927741
Thank you!
yours, LMR
Yes, it will always give a value such that nothing is above threshold. I'm not sure that the actual FDR threshold is defined in such cases.
doug
Lars M. Rimol wrote:
Hi,
Please disregard my previous mail. I was a little too trigger happy..., the 1.928 threshold set by tksurfer is too high for any vertex to survive so the two methods DO concur. I should actually look at the maps and not just the output...
The reason I asked is that I'm combining images to obtain an overall FDR, either for both hemispheres or for several comparisons. I don't see how this can be done in tksurfer so I'm using Nichols' function, which works just fine. I just don't know how to get an FDR threshold when no vertex has a p-value that survives - but tksurfer obviously produces such a threshold. How do I obtain that threshold?
(Or is it simply set just higher than the highest -log(p-value) in the image, for display purposes? It seems always to be just above.)
thanks (and apologies for the confusion), LMR
From: larilin@gmail.com To: freesurfer@nmr.mgh.harvard.edu Subject: fdr thresholds in tksurfer compared to Nichol's function Date: Thu, 25 Jun 2009 14:07:08 +0200
Hi,
I've used the fdr function in tksurfer and compared the results to Tom Nichol's matlab function (http://www.sph.umich.edu/~nichols/FDR/ http://www.sph.umich.edu/%7Enichols/FDR/). I use p = 10.^-abs(pfile), where "pfile" is 'sig.mgh'.
I find that the p-threshold (pID, assuming positive dependence) is usually exactly the same as the one in tksurfer (probably because you use the same algorithm..?).
But in some cases where pID is empty, which should mean that there is no suprathreshold vertex, I do get a threshold in tksurfer. Here's an example:
For one contrast where pID was returned empty in Nichol's matlab function, I got a threshold of 1.927741 in tksurfer.
Is this supposed to happen?
The output is:
MRISfdr2vwth(rate=0.05000, sign = 0, 1, only marked = 0) = 1.927741
surfer: MRISfdr2vwth with rate 0.05 and sign 0 returned threshold 1.927741
Thank you!
yours, LMR
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas N Greve -
Lars M. Rimol