[Mristudio-users] GE 15 gradient table

[susumu mori]

Yes,

There are three cases when things can go wrong even if the colormaps look
fine.

1) Simple mistake in the gradient table (e.g. [1, -1, 0] instead of [1, 1,
0] in one of the tables).

Here I assume you made a mistake in one or several lines of the gradient
table by typos. This is the most difficult kind of problems to catch.

When you do tensor calculation, it asks "Automated Rejection?". This
function is not working well in the current implementation. We are about to
release a much better version soon, but there is one handy function, called
"Create images of the DWI-Fitting" and "Create images of the difference
between ....". Click these two check boxes. For outlier rejection threshold,
put a large number to effectively turn off this function. After calculation,
click "Original-Mean-ADC-STD" button. Here, you can find raw DWI,
theoretical DWI back-calculated from fitting results, and difference of the
two. Think these results in this way; suppose you have 4 points in a X-Y
plot and did a linear fitting. After fitting, you can measure error between
the fitting results and each measured point. The bottom row of the
inspection window shows the difference (error) maps. These error maps should
look random noise. If there is any mistake in the gradient table, you can
immediately tell because you can clearly see anatomical structures in this
error map. Also, these error maps are handy to visually see the amount of
eddy current and subject motion.

2) Oblique angle

When you are using oblique angle axial imaging AND
> you are using Philips scanner AND using the "gradient overplus" option.
> you are using Siemens scanner AND the operating system prior to VB15

Then you need to re-calculate your gradient table based on the oblique
angles. If you are a Philips scanner user, you can get a handy software from
mri.kennedykrieger.org that can do the calculation. If you are using DICOM,
you can simply click "rotate gradient if applicable" beneath the gradient
table input section. Then the software can read the oblique info from the
header and do the recalculation

3) +/- sign

If +/- sign is wrong for only one of the table lines, it is the case
described in #1. If +/- sign of, for example, all X components of all lines
are flipped, that's a totally different story. There is a detailed
description of this issue in the manual of DtiStudio. Basically, +/-
definition of any given gradient table is arbitrary, and dynamically changed
depending on imaging protocol. If +/- signs of one of the X, Y, Z components
are flipped, you won't notice by looking at eigenvalue-based maps such as MD
and FA. The color map also looks correct. You notice the problem only when
you start fiber tracking. Please refer to the manual for detail.


Susumu

On Sun, Sep 27, 2009 at 1:31 PM, <miguel.burgaleta at uam.es> wrote:

>
>
> Hi Susumu,
>
> I'm sorry but what I was trying to mean is if I can trust the gradient
> table whose values come from the DICOM headers.
>
> I have 2 gradient tables: one from the manufacturer (medium GE) and one
> from the DICOM headers. When I use the former I get wrong colormaps (I have
> validated this with other software), but when I use
> the latter, the colormaps *seem* to be OK (red corpus callosum, symmetric
> colors).
>
> So the question is: even if the CC is solid red and the colormap is
> symmetric, is it still possible that the gradient table is wrong? Since I am
> a beginner I am not able to see it just by looking at the colormap.
>
> I know this is a basic question, perhaps beyond the scope of this list. So
> any help will be highly appreciated.
>
> Thanks,
> Miguel
>
>
>
> Quoting susumu mori <susumu at mri.jhu.edu>:
>
>  Hi Miguel,
>>
>> There are two possibilities;
>>
>> 1) your table is wrong.
>> 2) your table format is wrong.
>>
>> I suspect #2. The table has to be rewritten to DtiStudio format, which is;
>>
>> 0: x, y, z
>> 1: x, y, z
>>
>> Common mistakes are;
>>
>> 0; x, y, z (";" instead of ":")
>> 1: x y z (missing ",")
>> .
>> .
>> (there is no [0, 0, 0] in the table)
>>
>> When you have a long table, it is sometimes difficult to catch one
>> mistake.
>> In this case, please do the following;
>>
>>  Use your current table and load the images. You will get the problem you
>>>
>> reported.
>>
>>> Close the window and choose "DtiMapping" again.
>>> You will get the initial data input window, which still remembers all the
>>>
>> parameters you specified last time, including the gradient table.
>>
>>> Closely inspect the gradient table. This time it contains the table which
>>>
>> is NOT the one you typed in but the version interpreted by the software.
>> If
>> you made a mistake in the syntax, it is much easier to catch it in this
>> version.
>>
>> I hope this will help.
>>
>> Susumu
>>
>> On Fri, Sep 25, 2009 at 7:07 PM, <miguel.burgaleta at uam.es> wrote:
>>
>>
>>> Dear DTIStudio group,
>>>
>>> I have a beginner's question regarding the gradient table that I
>>> should use with my DTI. The scanner is a GE and I already have the
>>> standard medium table from the manufacturer, but I get weird FA
>>> colormaps when using it (non-symmetric, corpus callosum not red).
>>> However, if I use a gradient table made by reading the DICOM headers
>>> with dcm2nii, I get a gradient table that leads to symmetric, solid
>>> red CC FA colormaps.
>>>
>>> I have read in this list before that this is a good criterion to make
>>> sure that the gradient table is the right one, but also heard that it
>>> could be incorrect too! So the question is how to make sure that the
>>> gradient table is the right one?
>>>
>>> I appreciate your help.
>>>
>>> Miguel
>>>
>>>
>>> _______________________________________________
>>> 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/20091002/ad034eef/attachment-0001.html