[Insight-users] DICOM series

Fri Oct 22 09:22:40 EDT 2004

Uday, 

Glad the NumericSeriesFilenames worked. 

To interpolate slices, you can use the ResampleImageFilter.  You can
specify the type of iterpolation (NearestNeighbor, Linear, BSpline,
WindowedSinc) by setting an ImageFunction as the appropriate interpolator.

The ResampleImageFilter can seem complicated to use.  The ITK User's
Guide walks through a series of example uses for scaling images, rotating
images, changing resolutions.  In your instance, you'll want to set the 
transform to the identity and set the output size and output pixel spacing
so that you have the same size/spacing in xy as the original data and 
change the size/spacing in z so that you get interpolated slices.

Jim

-----Original Message-----
From: Uday Kurkure [mailto:udaykurkure at yahoo.com]
Sent: Thursday, October 21, 2004 5:17 PM
To: Blezek, Daniel J (Research); Miller, James V (Research);
insight-users at itk.org
Subject: RE: [Insight-users] DICOM series

Hello James and Daniel,

No, these images do not have Echo Time tag. 
I impemented the code to read the DICOM series using
NumericSeriesFilenames and it works great. I can use
it on both primary and secondary captures. 
Thanks for your valuable suggestions.

One more thing, is there any standard method to
perform interpolation to obtain pixels between the
slices. The interslice distance in my data is 10mm
which is greater than interpixel distance in a slice. 

-Uday

--- "Blezek, Daniel J (Research)" <blezek at crd.ge.com>
wrote:

> If these are multiple echo MR images (I couldn't
> tell from the post), you
> can sort on 0018,0081 (Echo Time) before
> SliceLocation.
> 
> -dan
> 
> -----Original Message-----
> From: Miller, James V (Research)
> [mailto:millerjv at crd.ge.com]
> Sent: Thursday, October 21, 2004 3:51 PM
> To: 'Uday Kurkure'; Miller, James V (Research);
> insight-users at itk.org
> Subject: RE: [Insight-users] DICOM series
> 
> 
> If the filenames correspond to the image numbers,
> i.e. filenames like
> IM100, IM101, etc so that 
> 
> IM100: location 1, time 1
> IM101: location 1, time 2
> IM102: location 1, time 3
> ...
> IM10x: location 2, time 1
> ...
> ...
> 
> then  you could use the NumericSeriesFileNames to
> generate filenames
> that can be passed to the ImageSeriesReader.  The
> NumericSeriesFileNames
> takes a start index and a increment and a file
> pattern.  This would 
> allow you to collect every Nth image into a series.
> 
> I wonder why there isn't a DICOM tag to identify the
> timestep uniquely.
> 
> Jim
> 
> 
> 
> -----Original Message-----
> From: Uday Kurkure [mailto:udaykurkure at yahoo.com]
> Sent: Thursday, October 21, 2004 1:42 PM
> To: Miller, James V (Research);
> insight-users at itk.org
> Subject: RE: [Insight-users] DICOM series
> 
> 
> 
> Hello James,
> 
> You were right. The Instance UIDs (and Image Numbers
> too) are 0 for all the secondary captures. So, I
> would
> prefer to sort the primary captures. About time
> tags,
> I know of two:
> 
> Acquisition time: 0008,0032 - The time the
> acquisition
> of data for the image started.
> 
> Image time: 0008,0033 - The time the image pixel
> creation started.
> 
> But these tags are same for all the images in a
> series.  So they are of no help to me.
> 
> When I sort a series by image number the data has
> the
> order: 
> location L1, time T1
> location L1, time T2
> ...
> location L1, time TN
> ...
> location L2, time T1
> location L2, time T2
> ...
> location L2, time TN
> ...
> ...
> location LM, time TN
> so on.
> 
> So one thing I can think of is to have user input M
> (number of slices in a volume) and N (number of
> phases), and group the slices in N groups 
> 
> - taking every Nth slice in one group - OPTION :
> EVERY
> N PHASE
> 
> - For data where the slices are ordered location
> wise
> first, we can just group first N slices in first
> group
> and so on - OPTION: DEFAULT
> 
> Do you think its possible? If yes, how can I do it
> in
> ITK? 
> 
> Thanks,
> 
> Uday
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> --- "Miller, James V (Research)"
> <millerjv at crd.ge.com>
> wrote:
> 
> > Uday, 
> > 
> > Do you know what DICOM tag distinguishes the time
> > instants? We can add
> > the ability to the DICOM code to group based on
> > time.
> > 
> > I have seen the second issue before. Can you check
> > the Instance 
> > UIDs (0x0008, 0x0018) for each of the images in
> the
> > new folder?  
> > 
> > You can use DICOMWorks (free download) to look at
> > the DICOM tags.  
> > I have seen secondary captures use the same
> Instance
> > UID for each image 
> > (say 0.0.0.0).  The UIDs are supposed to be unique
> > on an image to 
> > image basis.  The DICOM code in ITK uses the 
> UID's
> > as keys in several 
> > maps.  If the UIDs are not unique, the a single
> file
> > gets read for each 
> > slice in the volume. The short term fix is to fix
> > the UIDs in the DICOM
> > files.  DICOMWorks allows you to do this as well.
> > 
> > Check the DICOM tags (0x0008, 0x0018) to see if
> the
> > field is unique across
> > all your images.
> > 
> > Jim
> > 
> > 
> > 
> > -----Original Message-----
> > From: Uday Kurkure [mailto:udaykurkure at yahoo.com]
> > Sent: Wednesday, October 20, 2004 6:26 PM
> > To: insight-users at itk.org
> > Subject: [Insight-users] DICOM series
> > 
> > 
> > 
> > 
> > Hello all,
> > 
> > I have been playing with ITK DICOM series image
> > read/write and MRIconvert to write a volume image
> in
> > .raw format. The findings and problems:
> > 
> > 1. In a single series there are slices of volumes
> > acquired at different time instants. And
> > unfortunately, the acquisition time label in the
> > headers is same for all. So, sorting by slice
> > location
> > doesn't work properly as there are more than one
> > slices at the same position (of different time
> > instants). Is there any other way to sort the
> slices
> > in a volume and also sort the volumes?
> > (These are original primary capture images)
> > 
> > 2. I myself sorted the images for one volume
> > (acquired
> > at one time instant)and saved them in a separate
> > folder with naming convention: 
> > msPP_VV_SS.dcm, where PP is subject number, VV is
> 
=== message truncated ===

__________________________________
Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.
http://promotions.yahoo.com/new_mail