[IGSTK-Users] Multiframe DICOM

Thu Jan 6 03:03:11 EST 2011

Hi Ziv

Sorry about terms I used to describe the data-type - it's an XA projection
of the cardiac-chamber (phantom), single-plane fluoro.

Thanks for pointing us in the right direction. 

Cheers,

Satish

From: Ziv Yaniv [mailto:zivy at isis.georgetown.edu] 
Sent: Wednesday, January 05, 2011 5:23 PM
To: Satish
Cc: igstk-users at public.kitware.com
Subject: Re: [IGSTK-Users] Multiframe DICOM

Hi Satish,
"Yes it is a xa projection (axial slices from what I can see)."
Data type is still not clear. XA can be either x-ray projections or a
reconstruction, axial slices, it cannot be both. See
http://midas.kitware.com/community/view/34 for examples.

IGSTK wraps relevant parts of ITK and provides the added functionality
required for image-guided interventions. As it is dependent on ITK you
obviously always have ITK at your disposal when you are using IGSTK. 

Most likely approach to achieve what you want (image
segmentation/registration):
a. load the data using IGSTK reader.
b. perform the processing using an IGSTK component that you implement
(internally use ITK components + your custom components).
c. display using IGSTK components.

Data transfer: 
a-->b Get the ITK image from the reader and work on it.
b-->c Set the resulting ITK image into the relevant spatial object, take a
look at ImageReaderToImageSpatialObject (igstkImageReader.h) .  

Not clear what you mean by "DICOM fluoro directly from the c-arm,". If you
are talking about frame-grabbing the data then that is currently being
developed by Patrick as part of ITK and will be brought into IGSTK. If you
are talking about doing online DICOM transfer of the fluoro images then you
will need a DICOM client. This is not part of IGSTK and there is no plan to
provide this capability in IGSTK (DCMTK provides this type of service). 

With regard to video overlay, currently IGSTK does not do this, but users of
the toolkit have developed such  applications:
http://public.kitware.com/IGSTKWIKI/images/b/b1/Daly_CARS2010_IGSTKUserGroup
.pdf

                     regards
                              Ziv

On 1/5/2011 4:00 AM, Satish wrote: 

Hi Ziv,

Thanks for elaborating on this. Yes it is a xa projection (axial slices from
what I can see).

I followed your (and Louis and other) suggestion (added xa image reader
objects) and have created a new igstk app in visual studio 9.0 on the lines
of needle-biopsy app, and is now able

to load the image file and display in the 4 quadrants, only if they are
loaded from a folder with .dcm image series. Had to split the large DICOM
file into its frames as .dcm into 

a folder to do this. It was not loading if I tried with one single large
file. 

I am by no means a expert on igstk, and might me missing many things here.
Anyway, the goal is to load large xa modality files and perform segmentation
based on our custom

algorithms which intend to use ITK for this. Registration and visualization
will follow. 

First question is will igstk allow us to access full potential of ITK for
segmentation, or we have to rely on what igstk provides. Also in cases like
the hack you mentioned, how do we

Integrate ITK pipe into igstk if we have to fall back on ITK to load/process
large DICOM files through igstk.

Also along the roadmap is to be able to grab the DICOM fluoro directly from
the c-arm, and superimpose 3d data over 2d-fluoro, question is will igstk
support this kind of approach.

Again I'm not an expert on igstk (I'm .net architect), so are we on right
track?

Cheers

satish

From: Ziv Yaniv [mailto:zivy at isis.georgetown.edu] 
Sent: Tuesday, January 04, 2011 5:11 PM
To: igstk-users at public.kitware.com; Satish
Subject: Re: [IGSTK-Users] Multiframe DICOM

Hi Satish,

Multi-frame DICOM is a bit too generic. Are you talking about a set of
slices comprising a volume that are contained in a single file instead of
multiple single slice files? Perhaps these are the XA projection images
which you were previously asking about? 

Also, if the DICOM tag for modality  is not CT the igstkCTImageReader will
refuse to read it. For example a set of axial slices from cone-beam CT that
have an XA modality tag, even though they are equivalent to a CT. This is
intentional. 
If we expect a CT we don't want to allow the user to load an MR without the
developer knowing about this.

I suggest that you first try to read the data using an igstk reader. If non
of them work then try reading using an ITK reader. Once you establish which
pixel type and other parameters work within ITK you can implement an igstk
reader which will only accept that type. This ensures that you only read the
expected data type (important if you want preset window and level values to
work as expected, etc.).

If we are still talking about the XA projection images, then the following
"hack" in ITK may work for you; load the image set as a single 3D image.

This may not work if the movie is too long as ITK tries to load the whole
thing. 

If it does work you need to BE CAREFUL with your usage of this "volume". The
ITK concept of an image associates a pixel spacing with each dimension, thus
your 3D volume will most likely have a pixel dimension of 1 in each
direction. In the image, x-y, plane this is fine, but your "z" plane will be
time which should not be treated in the same way as space "x-y".
When/if you apply ITK filters to this type of data you will most likely need
to apply them in a "slice by slice" manner. 

                    hope this helps
                                       Ziv

On 1/4/2011 11:21 AM, Satish wrote: 

Hi all,

Is it possible to read a multi-frame DICOM file (approx 250MB) with the
igstkCTImageReader ?

Or do we have to split it into multiple .dcm files and read.if so how do we
split?

Regards

Satish Kumar B.G

Technical Architect

COMPUSOL

_______________________________________________
Powered by www.kitware.com

Visit other Kitware open-source projects at
http://www.kitware.com/opensource/opensource.html

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/cgi-bin/mailman/listinfo/igstk-users

-- 
Ziv Yaniv, PhD., Research Assistant Professor
Imaging Science and Information Systems (ISIS) Center
Department of Radiology
Georgetown University Medical Center
2115 Wisconsin Avenue, Suite 603
Washington, DC, 20007,

Phone: +1-202-6877286
Fax: +1-202-784-3479
email: zivy at isis.georgetown.edu
web: http://isiswiki.georgetown.edu/zivy/

-- 
Ziv Yaniv, PhD., Research Assistant Professor
Imaging Science and Information Systems (ISIS) Center
Department of Radiology
Georgetown University Medical Center
2115 Wisconsin Avenue, Suite 603
Washington, DC, 20007,

Phone: +1-202-6877286
Fax: +1-202-784-3479
email: zivy at isis.georgetown.edu
web: http://isiswiki.georgetown.edu/zivy/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/igstk-users/attachments/20110106/17cac871/attachment.html>