Hello,
Thank you both for your responses.
I've attached some figures to show the differences in results at p-values of 01 and 05 for versions r409 and r435 respecively.
I checked the CHANGES.txt file on the vbm8 page, and noticed some changes in between versions that could have an effect:
-Removed changes from r409 in GBM.m.
-Do not delete deformation field in longitudinal batch mode in cg_vbm8_longitudinal.m.
-Removed bug in longitudinal batch and deformation utility, where some images were not warped.
-Some changes in longitudinal batch (interpolation, bias regularization).
-Removed some typos in new realignment code.
I guess my concern is the marked differences in results, the visual inspection of the grey matter segmentions, resulting .txt files from relalignment, etc. Would modifications cause this much of a difference? And, which results should we be comfortable in accepting given this much disparity?
Michael
On Mon, 30 Jan 2012 09:21:01 +0100, Marko Wilke <[log in to unmask]> wrote:
>Hello Adam, Michael,
>
>just to follow this up, it is to be expected that an ever-improving
>piece of software such as vbm8 gives different results with different
>versions (else there weould be no sense in improving it :). If you want
>to know what changed, there is a very meticuous chages.txt file in the
>distribution. Irrespective of that, I think it is generally a really
>good idea to use one software revision for the (final) analysis of all
>subjects.
>
>Cheers,
>Marko
>
>Adam Gorka wrote:
>> Hello Michael,
>>
>> Our group has been collecting data for two years and we have been
>> processing it as we go. We noticed the same thing. As we collected data
>> across time, we updated our versions of vbm8. To make sure the new
>> versions were the same as the old, we processed the same data with both
>> versions. The different versions produced different Total Grey Matter
>> /White Matter/ CSF volumes in the text file generated during
>> preprocessing. Additionally, the grey matter segmented image produced
>> was different for the old and new versions.
>>
>> I am unsure what is changed across the different versions that produces
>> this difference. We chose to process all our data with the same version.
>> Hope this helps.
>>
>> Adam
>>
>>
>>
>> On Fri, Jan 27, 2012 at 1:52 PM, Michael <[log in to unmask]
>> <mailto:[log in to unmask]>> wrote:
>>
>> Hello,
>>
>> I've conducted some longitudinal vbm analysis back in spring 2011
>> with version vbm8_r409. The results looked promising. However, after
>> making the upgrade to vbm_r435, I noticed the same exact analysis i
>> conducted on the same data set yielded different results. Just to
>> check, I preprocessed the data set using version 409 and also using
>> 435. So, i'm assuming there is a difference in the preprocessing
>> calculations and this is due to the upgrade(s)? Has anybody else
>> experienced this? Your help would greatly be appreciated.
>>
>> Thanks,
>>
>> Michael
>>
>>
>
>--
>____________________________________________________
>PD Dr. med. Marko Wilke
> Facharzt f�r Kinder- und Jugendmedizin
> Leiter, Experimentelle P�diatrische Neurobildgebung
> Universit�ts-Kinderklinik
> Abt. III (Neurop�diatrie)
>
>
>Marko Wilke, MD, PhD
> Pediatrician
> Head, Experimental Pediatric Neuroimaging
> University Children's Hospital
> Dept. III (Pediatric Neurology)
>
>
>Hoppe-Seyler-Str. 1
> D - 72076 T�bingen, Germany
> Tel. +49 7071 29-83416
> Fax +49 7071 29-5473
> [log in to unmask]
>
> http://www.medizin.uni-tuebingen.de/kinder/epn
>____________________________________________________
|