[Mristudio-users] DTI Studio

[susumu]

Hi Alexi,

 

Thank you for the posting. You are right. It seems that AIR made the
registration worse. This is how the STD images and image inspection window
should be used; to check the registration quality as well as image quality.
If you have multiple repetition data, you can calculate the STD image
(actually subtracted image if you have only two repetitions).
Misregistration would show up as a ring around the brain as you found. If
you don't have repetition, you can also calculate discrepancies between
fitted (back-calculated from tensor fitting results) and raw DWIs. These
images can be obtained by choosing "Outlier Rejection = yes" when you do
tensor calculation and activate "create images of DWI-fitting from the
estimated tensor" and "create images of the difference between DWI-fitting
and original" boxes. The outlier rejection algorithm itself is still not
working perfectly (better version is coming out soon), but the "discrepancy"
images (raw minus fitted images) are handy to detect misregistration, eddy
current, and image corruption.

 

Now, your question is, why AIR made registration worse. I don't have a
definite answer, but the most probable reason is the wrong threshold value.
Unfortunately AIR is very sensitive to the threshold value (the default is
50). The threshold is used to erase the noise floor and avoid AIR to try to
match noises. In my experience AIR tends to work better with higher
thresholds. The value 50 usually works for Philips scanner, but Siemens
tends to have much higher values. So, 150-250 seem to work better (I don't
know which scanner you are using). If you get poor results with AIR, the
first thing I recommend is to try higher threshold values. Once you get a
right value, usually you don't have to change it unless you have a major
upgrade or switch the scanner manufacture.

 

Please try it and let us know the results.

 

Thanks,

 

Susumu

 

  _____  

From: mristudio-users-bounces at mristudio.org
[mailto:mristudio-users-bounces at mristudio.org] On Behalf Of Alexi Klimenko
Sent: Saturday, March 15, 2008 2:56 PM
To: mristudio-users at mristudio.org
Subject: [Mristudio-users] DTI Studio

 

Hi Everyone, 

I have a question about Dtistudio and image registration.
In the attached file you can find two standard deviations 
images, one of them is processed with AIR. 

I am leaning towards choosing the first series (stdA top row) as the 
second one show consistent ring-type intensities. Which one is suitable
for further processing? 

To my suprise, however, stdB bottow row, is the one processed with
AIR. Thank you. 

Alexi

  

  _____  

Instant message from any web browser! Try the new
<http://ca.messenger.yahoo.com/webmessengerpromo.php> Yahoo! Canada
Messenger for the Web BETA

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20080315/e3d65db7/attachment.html