Dear Freesurfer experts,
We investigated the relation between cortical thickness and performance on several cognitive tasks within a large group of Parkinson’s disease patients, but are slightly puzzled by the results. We obtained several, both in the vertex-wise analysis in qdec and in the SPSS analysis in the a-priori parcelated areas, negative correlations (i.e. better performance relates to a thinner cortical area) for all our neuropsychological tasks. These negative correlations were unexpected and we are having a hard time interpreting them. After correcting for multiple comparisons, all effects failed to reach the statistical threshold, rendering no results at all. When we compared our patient group on a structural level with healthy controls, we did find expected results that made sense.
We also ran the exact same analyses (same group, same data) in VBM and there found several positive task-related correlations in expected areas.
We noticed that in previous literature almost no studies investigated the relation between cortical thickness and cognitive task-performance. This made us wonder whether Freesurfer is suited for these correlation-based analyses, or is it better to be used in between group analyses? And do you have any idea on how to explain our negative correlations? Is it possible we might have done something wrong?
Many thanks,
Stan
Hi Stan
we have found positive correlations between performance (in this case CVLT) and thickness, so it is certainly possible. And negative correlations aren't necssarily false - you could certainly imagine that successful pruning for example could help performance. Have you visually inspected the surfaces for accuracy?
cheers Bruce
On Tue, 5 Feb 2013, Berg, S.F. van den wrote:
Dear Freesurfer experts,
We investigated the relation between cortical thickness and performance on several cognitive tasks within a large group of Parkinson?s disease patients, but are slightly puzzled by the results. We obtained several, both in the vertex-wise analysis in qdec and in the SPSS analysis in the a-priori parcelated areas, negative correlations (i.e. better performance relates to a thinner cortical area) for all our neuropsychological tasks. These negative correlations were unexpected and we are having a hard time interpreting them. After correcting for multiple comparisons, all effects failed to reach the statistical threshold, rendering no results at all. When we compared our patient group on a structural level with healthy controls, we did find expected results that made sense.
We also ran the exact same analyses (same group, same data) in VBM and there found several positive task-related correlations in expected areas.
We noticed that in previous literature almost no studies investigated the relation between cortical thickness and cognitive task-performance. This made us wonder whether Freesurfer is suited for these correlation-based analyses, or is it better to be used in between group analyses? And do you have any idea on how to explain our negative correlations? Is it possible we might have done something wrong?
Many thanks,
Stan
Hi Bruce,
I checked the surface of every subject with tksurfer, inaccuracy's were founded in 13 of 140 subjects. These subjects were manually edited with control points and pial edits. Then i start recon-all on the subjects with edits again. After these procedure tksurfer showed a correct segmentation. Is that the correct procedure for checking inaccuracy's? Or do i need to check more?
http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/ControlPoints
http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/PialEdits
Many thanks,
Stan
________________________________________ From: Bruce Fischl [fischl@nmr.mgh.harvard.edu] Sent: 05 February 2013 15:54 To: Berg, S.F. van den Cc: Freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Difficult results from our PD data
Hi Stan
we have found positive correlations between performance (in this case CVLT) and thickness, so it is certainly possible. And negative correlations aren't necssarily false - you could certainly imagine that successful pruning for example could help performance. Have you visually inspected the surfaces for accuracy?
cheers Bruce
On Tue, 5 Feb 2013, Berg, S.F. van den wrote:
Dear Freesurfer experts,
We investigated the relation between cortical thickness and performance on several cognitive tasks within a large group of Parkinson?s disease patients, but are slightly puzzled by the results. We obtained several, both in the vertex-wise analysis in qdec and in the SPSS analysis in the a-priori parcelated areas, negative correlations (i.e. better performance relates to a thinner cortical area) for all our neuropsychological tasks. These negative correlations were unexpected and we are having a hard time interpreting them. After correcting for multiple comparisons, all effects failed to reach the statistical threshold, rendering no results at all. When we compared our patient group on a structural level with healthy controls, we did find expected results that made sense.
We also ran the exact same analyses (same group, same data) in VBM and there found several positive task-related correlations in expected areas.
We noticed that in previous literature almost no studies investigated the relation between cortical thickness and cognitive task-performance. This made us wonder whether Freesurfer is suited for these correlation-based analyses, or is it better to be used in between group analyses? And do you have any idea on how to explain our negative correlations? Is it possible we might have done something wrong?
Many thanks,
Stan
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
Hi Stan, the VBM results may be picking up changes in GM volume. These volume changes may be driven by differences in surface area and would not necessarily show up in a thickness analysis. Maybe the surface area is increasing with performance but the thickness is decreasing. You can do a surface-based analysis of surface area and/or volume (this is similar to a VBM analysis but on the surface). You can do this by specifying --meas area or --meas volume to mris_preproc. Make sure you have the mris_preproc patch (or are using 5.2).
doug
On 02/05/2013 09:54 AM, Bruce Fischl wrote:
Hi Stan
we have found positive correlations between performance (in this case CVLT) and thickness, so it is certainly possible. And negative correlations aren't necssarily false - you could certainly imagine that successful pruning for example could help performance. Have you visually inspected the surfaces for accuracy?
cheers Bruce
On Tue, 5 Feb 2013, Berg, S.F. van den wrote:
Dear Freesurfer experts,
We investigated the relation between cortical thickness and performance on several cognitive tasks within a large group of Parkinson?s disease patients, but are slightly puzzled by the results. We obtained several, both in the vertex-wise analysis in qdec and in the SPSS analysis in the a-priori parcelated areas, negative correlations (i.e. better performance relates to a thinner cortical area) for all our neuropsychological tasks. These negative correlations were unexpected and we are having a hard time interpreting them. After correcting for multiple comparisons, all effects failed to reach the statistical threshold, rendering no results at all. When we compared our patient group on a structural level with healthy controls, we did find expected results that made sense.
We also ran the exact same analyses (same group, same data) in VBM and there found several positive task-related correlations in expected areas.
We noticed that in previous literature almost no studies investigated the relation between cortical thickness and cognitive task-performance. This made us wonder whether Freesurfer is suited for these correlation-based analyses, or is it better to be used in between group analyses? And do you have any idea on how to explain our negative correlations? Is it possible we might have done something wrong?
Many thanks,
Stan
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Berg, S.F. van den -
Bruce Fischl -
Douglas N Greve