A beta version of the new VBM8 toolbox can be downloaded at the download section.
New features:
It was brought to my attention (thank you, Neil Ranjan Chatterjee) that data values above 1.0 were truncated after being processed with the vbm5 toolbox. Investigating the source of this truncation, I discovered that it only occurs after installation of the spm5 update 1048 or later. Update 1048 was released on Jan 8, 2008, and in it, the function spm_write_vol.m was modified. Before this update, a scale factor was estimated such that the maximum range of integer values was used. After the update, a truncation of integer data may occur if the scaling factor is pre-defined. Unfortunately, I have used a fixed scaling factor for modulated data and uint8 as the data type in the vbm5 toolbox. This worked fine before the update, but might lead to a truncation after the update was installed.
Regrettably, the truncation might also affect the statistical analysis of these data. If you have updated spm5 to version 1048 or higher (or after Jan 8, 2008), please check your modulated data. This tool will check your data for a truncation above 1.0.
If your data was truncated, you do not have to estimate the segmentation again, but you need to rewrite the data with the new vbm5 version 1.18 (use option “Write already estimated segmentations”).
A beta version of the new VBM5.1 toolbox can be downloaded at the download section.
If you tried to process multiple files using the option “VBM: Write segmentations only” the tool crashed. I have removed this bug and the new version 1.03 is available at the download section.
There was a bug in the longitudinal segmentation of VBM2, which caused a crash if you tried to segment more than two subjects.
Furthermore, the CSF images in cross-sectional and longitudinal segmentation were not masked which is neccessary for the HMRF approach. This resulted in classifying noisy regions outside the brain as CSF. Gray and white matter images were not affected.
It is recommended to update VBM2 to version 1.05 available at the download section.
The HMRF calculation was still too memory demanding. I have rewritten the HMRF algorithms for VBM2 and VBM5 to save memory. For MR images with 1 mm spatial resolution you need at least 512 MB RAM.
The new versions 1.03 of VBM2 and 1.01 of VBM5 can be downloaded at the download section.
The HMRF calculation for the first segmentation step of raw images was too memory hungry. I have removed this first HMRF computation which should be without impact on the segmentation results. The final HMRF step of the normalized data is still included. Furthermore, I have added an option to the “create customized template” function to delete temporary files after template computation.
The new version 1.02 of VBM2 can be downloaded at the download section.
The VBM toolboxes are a collection of extensions to the segmentation algorithm of SPM2 and SPM5 (Wellcome Department of Cognitive Neurology) to provide voxel-based morphometry (VBM). The toolboxes are named according to the SPM version. The software was developed by Christian Gaser (University of Jena, Department of Psychiatry) and is available to the scientific community under the terms of the GNU General Public License.
You are currently browsing the Structural Brain Mapping Group weblog archives for the VBM category.
Longer entries are truncated. Click the headline of an entry to read it in its entirety.
