External Email - Use Caution
Hi everyone,
I reviewed some of the other answers on the mailing list to compute a Cohen's d value following the sequence of commands below for a Qdec Freesurfer analysis:
1) cd glmdir/contrast fscalc gamma.mgh div ../rstd.mgh -o cohensd.mgh
2) mri_segstats --i cohensd.mgh --seg cluster.sig.ocn.mgh --exclude 0 --sum test.output
However, my test.output file looks like:
# ColHeaders Index SegId NVoxels Volume_mm3 StructName Mean StdDev Min Max Range
1 1 2398 2398.0 Seg0001 4.9611 0.8313 3.6525 7.3172 3.6647
2 2 5893 5893.0 Seg0002 5.1281 1.1535 3.6520 8.8040 5.1520
Why are the mean Cohen's d values (4.9611, 5.1281) so large? Am I interpreting these incorrectly?
Many thanks in advance for your help, Josh
You have created a circularity that is classic in neuroimaging (see Vul's paper on "VooDoo Correlations"). By using only voxels inside a cluster, you have selected voxels that are sure to show hi D values
On 9/16/20 7:48 PM, Alejandro Meruelo wrote:
External Email - Use CautionHi everyone,
I reviewed some of the other answers on the mailing list to compute a Cohen's d value following the sequence of commands below for a Qdec Freesurfer analysis:
cd glmdir/contrast fscalc gamma.mgh div ../rstd.mgh -o cohensd.mgh
mri_segstats --i cohensd.mgh --seg cluster.sig.ocn.mgh --exclude 0
--sum test.output
However, my test.output file looks like:
# ColHeaders Index SegId NVoxels Volume_mm3 StructName Mean StdDev Min Max Range
1 1 2398 2398.0 Seg0001 4.9611 0.8313 3.6525 7.3172 3.6647
2 2 5893 5893.0 Seg0002 5.1281 1.1535 3.6520 8.8040 5.1520
Why are the mean Cohen's d values (4.9611, 5.1281) so large? Am I interpreting these incorrectly?
Many thanks in advance for your help, Josh
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
External Email - Use Caution
Thank you for clarifying this confusing situation for me. I am attempting to compute Cohen's d values for regions found to be significant following Qdec group analysis using Monte Carlo simulation while controlling for multiple comparisons. I apologize for my ignorance, but is there a more accurate way to compute Cohen's d values for the regions identified using Monte Carlo simulation?
On Fri, Sep 18, 2020 at 8:03 AM Douglas Greve dgreve@mgh.harvard.edu wrote:
You have created a circularity that is classic in neuroimaging (see Vul's paper on "VooDoo Correlations"). By using only voxels inside a cluster, you have selected voxels that are sure to show hi D values
On 9/16/20 7:48 PM, Alejandro Meruelo wrote:
External Email - Use CautionHi everyone,
I reviewed some of the other answers on the mailing list to compute a Cohen's d value following the sequence of commands below for a Qdec Freesurfer analysis:
cd glmdir/contrast fscalc gamma.mgh div ../rstd.mgh -o cohensd.mgh
mri_segstats --i cohensd.mgh --seg cluster.sig.ocn.mgh --exclude 0
--sum test.output
However, my test.output file looks like:
# ColHeaders Index SegId NVoxels Volume_mm3 StructName Mean StdDev Min Max Range
1 1 2398 2398.0 Seg0001 4.9611 0.8313 3.6525 7.3172 3.6647
2 2 5893 5893.0 Seg0002 5.1281 1.1535 3.6520 8.8040 5.1520
Why are the mean Cohen's d values (4.9611, 5.1281) so large? Am I interpreting these incorrectly?
Many thanks in advance for your help, Josh
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
No, not unless you have another data set or you split your data into two sets. Using one set, you get the clusters, then compute the D from the other data set.
On 9/18/2020 3:16 PM, Alejandro Meruelo wrote:
External Email - Use CautionThank you for clarifying this confusing situation for me. I am attempting to compute Cohen's d values for regions found to be significant following Qdec group analysis using Monte Carlo simulation while controlling for multiple comparisons. I apologize for my ignorance, but is there a more accurate way to compute Cohen's d values for the regions identified using Monte Carlo simulation?
On Fri, Sep 18, 2020 at 8:03 AM Douglas Greve dgreve@mgh.harvard.edu wrote:
You have created a circularity that is classic in neuroimaging (see Vul's paper on "VooDoo Correlations"). By using only voxels inside a cluster, you have selected voxels that are sure to show hi D values
On 9/16/20 7:48 PM, Alejandro Meruelo wrote:
External Email - Use CautionHi everyone,
I reviewed some of the other answers on the mailing list to compute a Cohen's d value following the sequence of commands below for a Qdec Freesurfer analysis:
cd glmdir/contrast fscalc gamma.mgh div ../rstd.mgh -o cohensd.mgh
mri_segstats --i cohensd.mgh --seg cluster.sig.ocn.mgh --exclude 0
--sum test.output
However, my test.output file looks like:
# ColHeaders Index SegId NVoxels Volume_mm3 StructName Mean StdDev Min Max Range
1 1 2398 2398.0 Seg0001 4.9611 0.8313 3.6525 7.3172 3.6647 2 2 5893 5893.0 Seg0002 5.1281 1.1535 3.6520 8.8040 5.1520Why are the mean Cohen's d values (4.9611, 5.1281) so large? Am I interpreting these incorrectly?
Many thanks in advance for your help, Josh
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
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Alejandro Meruelo -
Douglas Greve -
Douglas N. Greve