Dear Min,
I agree that the lateral parts of the thalamus and the s. nigra are poorly segmented. However, this is due to the fact that the intensity in these areas is very close to the intensity of WM. Although the contrast to the surrounding WM is obvious by eye it is hard for the algorithm to clearly differentiate between the GM areas that are too bright and the surrounding WM. Approaches based on prior probability maps like New Segment have even more difficulties for these areas. Unfortunately, I don't have any other solution for you...
Regards,
Christian
____________________________________________________________________________
Christian Gaser, Ph.D.
Departments of Psychiatry and Neurology
Friedrich-Schiller-University of Jena
Jahnstrasse 3, D-07743 Jena, Germany
Tel: ++49-3641-934752 Fax: ++49-3641-934755
e-mail: [log in to unmask]
http://dbm.neuro.uni-jena.de
On Wed, 13 Apr 2011 12:34:18 -0600, Min Liu <[log in to unmask]> wrote:
>Dear Dr.Gaser,
>
>Thank you very much for taking time answering my question and pinpoint the
>crux. Very helpful.
>
>I went back to check the bias corrected T1-images. Although theoretically
>understand what you mean by 'too bright', practically I still need a pair of
>good eyes to judge my data. Could you please take a look at the graph I
>created out of my T1-image with both bias corrected maps and segmentation
>results at the level of substantia nigra and thalamus. Do you think VBM8 has
>done a good job? In my opinion, the left side substantia nigra was poorly
>defined due to lack of contrast at that region and both side thalami were
>limited only to the central part. I might need to do more bias correction.
>
>Thank you again!
>Sincerely,
>Min
>
>On Mon, Apr 11, 2011 at 7:03 AM, Christian Gaser <
>[log in to unmask]> wrote:
>
>> Dear Min,
>>
>> the partial volume estimation (PVE) approach should have some advantages
>> esp. in the basal ganglia or thalamus. These structures will reveal PVE
>> segmentation values which are in the middle between pure gray and white
>> matter segmentations. Typical values are around 0.5 for both gray as well as
>> white matter. However, sometimes bias correction might fail (esp. or
>> scanners with >=3T) and these structures will be identified as white matter.
>> This can be checked by displaying the bias corrected T1-image, where these
>> central structures are too bright. In this case I recommend to modify the
>> settings for bias correction (e.g. bias regularization and/or fwhm).
>>
>> Regards,
>>
>> Christian
>>
>>
>> ____________________________________________________________________________
>>
>> Christian Gaser, Ph.D.
>> Departments of Psychiatry and Neurology
>> Friedrich-Schiller-University of Jena
>> Jahnstrasse 3, D-07743 Jena, Germany
>> Tel: ++49-3641-934752 Fax: ++49-3641-934755
>> e-mail: [log in to unmask]
>> http://dbm.neuro.uni-jena.de
>>
>>
>> On Thu, 31 Mar 2011 15:42:09 -0600, Min Liu <[log in to unmask]> wrote:
>>
>> >Dear all,
>> >
>> >It comes to me that using VBM8 toolbox has an advantage over the common
>> SPM
>> >routine introduced by the 'Partial Volume Estimation' function. That means
>> >it can better segment GM-WM mixed classes such as thalamus and other
>> >subcortical structures. It turns out that most part of the thalamus and
>> >subcortical grey matter are designated as white matter as shown in the
>> graph
>> >I appended to this email. I am currently doing a VBM on white matter
>> between
>> >a cohort of patients and normal subjects. Significantly reduced WM volume
>> >was shown in red and overlapped on a representative WM segmentation map. I
>> >am wondering how much I can trust the result. The detected change in
>> yellow
>> >box consist of thalamus, putaman, pallidum and caudate, those regions that
>> >shouldn't be classified as white matter. The detected change in green box
>> >seems nail some true white matter. Can I say change in green box reliable
>> >and change in yellow box not? Thank you for your comments in advance!
>> >
>> >Min
>> >
>>
>>
>>
>
|