Mixing datasets

[Luke Whitmore]

Hi

I have three sets of DICOM files, each of which provide MRI images of the same body part along three main axes.

Is it possible to combine DIMCOM dataset in the same world space using the slice3dVWR module to increase the resolution of data?

Thanks for a great tool

Luke 

[Charl Botha]

Dear Luke,

On Thu, Sep 6, 2012 at 5:42 PM, Luke Whitmore <lwhi...@gmail.com> wrote:

I have three sets of DICOM files, each of which provide MRI images of the same body part along three main axes.

Is it possible to combine DIMCOM dataset in the same world space using the slice3dVWR module to increase the resolution of data?

Francois Malan has just recently developed and extensively tested a DeVIDE approach for doing exactly this. It involves the DICOMAligner module, but I'm not sure which other module. Francois, could you help out?

 

Thanks for a great tool

Our pleasure!

See you,

Charll/

[Charl Botha]

---------- Forwarded message ----------
From: Francois Malan <fma...@medvis.org>
Date: Fri, Sep 7, 2012 at 2:30 PM
Subject: Re: Mixing datasets
To: Charl Botha <c.p....@tudelft.nl>
Cc: devide...@googlegroups.com


Dear Luke

Charl mentioned the correct module - it is called DICOMAligner and is
shipped with DeVIDE. But this is only the first step in actually
merging the images.
I didn't make a single module for the whole process - instead I can give you
an example network that merges 3 DICOM image volumes. From this
exampple you can see how it works and adjust it for processing more
sets as required.

Basically the process works as follows:
1) Using DICOMAligner, align each DICOM image volume to the
scanner/world/patient LPH coordinate frame (as opposed to the images'
coordinate frame).
2) Decide what the output image's extent and spacing should be. This
is done in the CodeRunner - automatically determined from the
DICOMAligner outputs,
    but can be manually overridden as required
3) Resample the aligned image volumes to the output volume's grid
(using probe filters)
4) Average the resampled image volumes using ImageMathematics
5) Cast to correct bit depth (intermediate steps are performed as
floats, final often must be short)
6) Write to disk

Regards
Francois

--
Francois Malan, M.Sc.
Researcher
LUMC Leiden (Orthopaedics) /
TU Delft (Graphics & Visualization)
Mobile: +31 62 493 0748
E-mail: fma...@medvis.org
http://graphics.tudelft.nl/People/FrancoisMalan

--
dr. charl p. botha :: http://graphics.tudelft.nl/CharlBotha
work: c.p....@tudelft.nl personal: cpb...@cpbotha.net