Hello Surfers,
I am trying to verify the p-value sig.mgh maps outputted from the mri_glmfit command as the first part of a more advanced analysis. We have been extracting the thickness data for a single vertex from each of our 58 subjects in two control groups and then running a significiance analysis is SAS. We thought that the output from SAS should match the p-value in the sig.mgh file, but the p-values do not match at all. We have tested multiple data points. Does freesurfer do some sort of a more advanced significance analysis that is not in the standard documentation?
Thank you very much, Jeff Sadino
_________________________________________________________________ Windows Live™ SkyDrive™: Get 25 GB of free online storage. http://windowslive.com/online/skydrive?ocid=TXT_TAGLM_WL_SD_25GB_062009
Jeff,
sig.mgh actually stores the -log10(p), not p, so to get the true p-value of say, -2 (in the sig.mgh), take 10^-2 (giving you the true p-value of 0.01).
Nick
On Mon, 2009-06-08 at 16:48 -0600, Jeff Sadino wrote:
Hello Surfers,
I am trying to verify the p-value sig.mgh maps outputted from the mri_glmfit command as the first part of a more advanced analysis. We have been extracting the thickness data for a single vertex from each of our 58 subjects in two control groups and then running a significiance analysis is SAS. We thought that the output from SAS should match the p-value in the sig.mgh file, but the p-values do not match at all. We have tested multiple data points. Does freesurfer do some sort of a more advanced significance analysis that is not in the standard documentation?
Thank you very much, Jeff Sadino
Windows Live™ SkyDrive™: Get 25 GB of free online storage. Get it on your BlackBerry or iPhone. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Also, p-values for t-tests are two-sided.
Nick Schmansky wrote:
Jeff,
sig.mgh actually stores the -log10(p), not p, so to get the true p-value of say, -2 (in the sig.mgh), take 10^-2 (giving you the true p-value of 0.01).
Nick
On Mon, 2009-06-08 at 16:48 -0600, Jeff Sadino wrote:
Hello Surfers,
I am trying to verify the p-value sig.mgh maps outputted from the mri_glmfit command as the first part of a more advanced analysis. We have been extracting the thickness data for a single vertex from each of our 58 subjects in two control groups and then running a significiance analysis is SAS. We thought that the output from SAS should match the p-value in the sig.mgh file, but the p-values do not match at all. We have tested multiple data points. Does freesurfer do some sort of a more advanced significance analysis that is not in the standard documentation?
Thank you very much, Jeff Sadino
Windows Live™ SkyDrive™: Get 25 GB of free online storage. Get it on your BlackBerry or iPhone. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hello Everyone, Thank you all for your replies. We converted the values into -log10(p) and accounted for two-tailed, but the values were still off, and they are not even off in any consistent pattern. For example: FS: .1541 .1109 .0749 .0641 .0514 .0281 .0182 .0141 .0132 .0025 .0007 .0006 SAS: .3226 .0852 .2131 .0818 .0760 .2425 .6196 .0020 .0458 .0039 .0003 .0020
Our SAS program only gives output to four digits. Could the discrepencies be coming from rounding errors?
Mahalo, Jeff
Date: Mon, 8 Jun 2009 19:21:53 -0400 From: greve@nmr.mgh.harvard.edu To: nicks@nmr.mgh.harvard.edu CC: jsadino@hotmail.com; freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] mri_glmfit significiance verification outside freesurfer
Also, p-values for t-tests are two-sided.
Nick Schmansky wrote:
Jeff,
sig.mgh actually stores the -log10(p), not p, so to get the true p-value of say, -2 (in the sig.mgh), take 10^-2 (giving you the true p-value of 0.01).
Nick
On Mon, 2009-06-08 at 16:48 -0600, Jeff Sadino wrote:
Hello Surfers,
I am trying to verify the p-value sig.mgh maps outputted from the mri_glmfit command as the first part of a more advanced analysis. We have been extracting the thickness data for a single vertex from each of our 58 subjects in two control groups and then running a significiance analysis is SAS. We thought that the output from SAS should match the p-value in the sig.mgh file, but the p-values do not match at all. We have tested multiple data points. Does freesurfer do some sort of a more advanced significance analysis that is not in the standard documentation?
Thank you very much, Jeff Sadino
Windows Live™ SkyDrive™: Get 25 GB of free online storage. Get it on your BlackBerry or iPhone. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
In order to help us help you, please follow the steps in: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
_________________________________________________________________ Insert movie times and more without leaving Hotmail®. http://windowslive.com/Tutorial/Hotmail/QuickAdd?ocid=TXT_TAGLM_WL_HM_Tutori...
Those differences are much too large to be rounding errors. Are you sure that you are running identical models in mris_glmfit and SAS, and that the data that you are testing in SAS is in fact what is being processed in mris_glmfit (e.g., same degree of smoothing)?
-Mike H.
Hello Everyone, Thank you all for your replies. We converted the values into -log10(p) and accounted for two-tailed, but the values were still off, and they are not even off in any consistent pattern. For example: FS: .1541 .1109 .0749 .0641 .0514 .0281 .0182 .0141 .0132 .0025 .0007 .0006 SAS: .3226 .0852 .2131 .0818 .0760 .2425 .6196 .0020 .0458 .0039 .0003 .0020
Our SAS program only gives output to four digits. Could the discrepencies be coming from rounding errors?
Mahalo, Jeff
Date: Mon, 8 Jun 2009 19:21:53 -0400 From: greve@nmr.mgh.harvard.edu To: nicks@nmr.mgh.harvard.edu CC: jsadino@hotmail.com; freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] mri_glmfit significiance verification outside freesurfer
Also, p-values for t-tests are two-sided.
Nick Schmansky wrote:
Jeff,
sig.mgh actually stores the -log10(p), not p, so to get the true
p-value
of say, -2 (in the sig.mgh), take 10^-2 (giving you the true p-value
of
0.01).
Nick
On Mon, 2009-06-08 at 16:48 -0600, Jeff Sadino wrote:
Hello Surfers,
I am trying to verify the p-value sig.mgh maps outputted from the mri_glmfit command as the first part of a more advanced analysis. We have been extracting the thickness data for a single vertex from each of our 58 subjects in two control groups and then running a significiance analysis is SAS. We thought that the output from SAS should match the p-value in the sig.mgh file, but the p-values do not match at all. We have tested multiple data points. Does freesurfer do some sort of a more advanced significance analysis that is not in the standard documentation?
Thank you very much, Jeff Sadino
Windows Live SkyDrive: Get 25 GB of free online storage. Get it on your BlackBerry or iPhone. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
In order to help us help you, please follow the steps in: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
Insert movie times and more without leaving Hotmail®. http://windowslive.com/Tutorial/Hotmail/QuickAdd?ocid=TXT_TAGLM_WL_HM_Tutori... 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 -
Jeff Sadino -
Michael Harms -
Nick Schmansky