[Mristudio-users] How to combine DTI data from different runs

[susumu mori]

> - I loaded all 3 repeated scans by adding the folders to the Dicom
> Image File Folders, and all the scans loaded properly.
> - My guess is the next step is to realign all the images using AIR. I
> have a couple of questions about this.
>   - What is the optional upper bound?
>

AIR requires us to define a threshold to remove the background signals.
Otherwise, it tries to register the noise floor too. Also, AIR is sensitive
to the threshold. We usually do not use upper bound, but this is to remove
the CSF signals in b0 image to make the contrast closer to DWIs.

Because you have 3 repeated images, I recommend you to calculate STD of the
repeated scan using one of the buttons in the "DtiMapping" tab ("Mean, STD,
etc"). Then click "Original-ADC-Mean-STD" button. You can find two extra
rows of images; one is the average of the 3 repeated scans and the other is
the STD. Usually the STD images should look like nothing but noise. If the
subject moved, you can see clear anatomy due to image shift and thus a king
of edge-enhanced STD images.

You have to monitor what AIR does using this window. If the STD images look
like noise, you don't have to do AIR but because AIR smooths the images (and
thus enhance SNR), you'd better do AIR for all data. Make sure that AIR
don't make such a data worse. If you find mis-registration in the STD
images, AIR should correct them.

AIR results are sensitive to threshold and also cost function. It is purely
try & error. Usually once it works, you don't have to change often. My
experience says, threshold tends to be high side for robust performance.


>   - Which cost function is best to use?
>   - Should I adjust the gradients for the diffusion weighted images?
>

I don't use this function. If the subject moves so much that this gradient
correction based on image registration makes a significant impact on your
data (namely, if AIR rotates the brain by 5 degree, the gradient vector also
has to be rotated by 5 degree),  you'd better worry so many other things
about the validity of the data itself. However, it is true that some
reviewers believe in this and ask you "did you do it?"

>
> - Once Air finishes I have a bunch of images labeled Air-series-blah\
> -1, Air-series-blah\ -2, etc.
>  - How do I save these images so I can just reload them later and not
> have to redo-air.
>  - How do I get DTI studio to use these images rather than the
> original images for the FA calculation?
>  - How do I tell DTIstudio to combine the images across the repeats?
>

If you unclick "create new image after registration", you registered images
overwirte you original data and subsequent calculation will be based on the
registered images.

if you leave it checked, you get an extra dataset called "AIR_xxx". In this
case, you need to save the registered "AIR_xxx" images and reload them for
tensor calculation. For this, you use the "save" icon in the "Image" tab.
Choose the entire data (b0 + DWIs) and save them into 1 file (if you have
three repetitions, into 3 files) as "raw" format. This is so-called REC
format in DtiStudio.

When you have 3 repeated scans and load all of them, there is no particular
steps to combine them. DtiStudio uses all 3 scans to fit one tensor.



> Thanks again,
> Darren
>
>
>
> > Hi Darren,
> >
> > In a previous communication in the mailing list, I explained why we can
> not
> > use regular k-space averaging for DTI and, therefore, we average data
> after
> > magnitude calculation (we average images, not k-space data). If anybody
> > remembers which thread it was, I would appreciate it.
> >
> > Anyway, because we first have to calculate image before averaging, there
> is
> > no point to ask scanners to do the averaging. One downside of the
> > averaging-by-the-scanner is, if the images are mis-registered or there
> are
> > corrupted images, we can not undo the averaging.
> >
> > I believe that is why Siemens always save all images separately (without
> > creating an "averaged image" even if multiple averaging is specified.
> >
> > Please note the difference between "repeated scans (you submitted the
> same
> > protocol twice)" and "averaged scans (you specified 2 signal averaging)"
> > when you use Siemens. Either way, all images are saved individually and
> they
> > are basically the same studies. However, in the former case, the
> "repeated
> > images" are stored in different directories and Siemens console can not
> > combine them for DTI  calculation. With the latter case, the repeated
> images
> > are stored in the same directory and Siemens console can combine them to
> > calculate one DTI data.
> >
> > Either way, when you use DtiStudio, you need to load all files at once
> and
> > DtiStudio can combine them to calculate one DTI.
> >
> > Now going back to your question;
> >
> > 1) You can specify the number of signal averaging in Siemens. I'm not
> sure
> > why you thought you couldn't. However, if you meant, "Siemens saved all
> > images separately without averaging", then you are correct.
> > 2) DtiStudio can read repeated scans and combine them for one DTI
> > calculation.
> >
> > On Sun, Feb 28, 2010 at 7:54 PM, Darren Gitelman <
> > d-gitelman at northwestern.edu> wrote:
> >
> > > Hello:
> > >
> > > Is there a way to combine data across several repeats of a DTI
> sequence?
> > >
> > > The reason I ask is that we have a Siemens Trio scanner but are
> > > working with another group that has a GE scanner. On the GE scanner
> > > the DTI sequence allows NEX > 1 (they are getting 4), which appears to
> > > improve their SNR considerably. The Siemens sequences do not allow
> > > this, so we are repeating the DTI sequence several times, and want to
> > > combine the data across runs. Can this be done within DTIstudio or
> > > some other program? Would it be best to do this at the level of the
> > > DTI images or the FA maps?
> > >
> > > Thanks,
> > > Darren
> _______________________________________________
> Mristudio-users mailing list
> Mristudio-users at mristudio.org
> http://lists.mristudio.org/mailman/listinfo/
> Unsubscribe, send a blank email to:
> Mristudio-users-unsubscribe at mristudio.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20100301/963d3321/attachment.html