Hi Mitchell,

Answer to Q 2. "I believe that Modality 0008 0060 is a fixed value for
an entire
series. Where do I find this in the standard as well? "

Please see the section C.7.3.1.1.1 on Page 285 of the DICOM Standard
2007 version PDF no 3.
I hope it is the same what u are looking for.

About other questions I am searching for information and will let you
know if there is any finding...
Have a nice time!

Regards,
Prashant




On Feb 6, 7:42*am, shul wrote:
> Hi I have a simple question for Dicom Series;
>
> 1. Do all images in an dicom series *have to have same number of rows
> and columns (0028 0010 and *0028 0011)?
> Not that I have found any studies with different numbers of rows and
> columns in a single series, but it would make it easier to display a
> series- I can just check a single image instead *of all 1000
> images. .
>
> If so, where in the standard is this discussed?
>
> 2. I believe that Modality 0008 0060 is a fixed value for an entire
> series. Where do I find this in the standard as well?
>
> I recall someone once pointed out to me a good correction item (CP
> section of *David Clunie website) that was very helpful in getting a
> grip on extracting *thedicom *image data from common *CT and MRI
> formats. Can someone remind *me which one it is? It was easier to read
> than wading through sections 5 and 3.
> Perhaps it was CP014: or is there another one as well?
>
> Mitchell

In general, the images in a series do not have to have the same number
of rows and columns. There is an exception for PET, where Rows and
Columns must be consistent across the series. (There may be other
exceptions I've missed, too.)

Whether it's common or not for rows and columns to change within a
series is another issue. I'm just answering about what the spec says.

Here is a way to find this in the spec:

Look in PS 3.3 at Figure A.1-1 DICOM Composite Instance IOD
Information Model, which shows the relationships of DICOM Information
Entities (IE) in Composite Information Objects, including images.
Just below it you'll also find additional descriptions in A.1.2.3
SERIES IE and A.1.2.6 IMAGE IE. You'll see that one Series IE can
contain zero, one or more (0-n) Images IEs. Unless told otherwise,
like with PET, attributes in the Image IE can change within a series
and any attribute in the Series IE must be the same for all of the
images (and other things) in a series.

Now, you need to find what IE has Rows (0028,0010) and Columns
(0028,0011). From the descriptions above, you'd probably guess that
they are in the Image IE. You'd be right, but let's prove it.

For images (not print boxes, directory entries, SR's, GSPS', etc.),
Rows (0028,0010) and Columns (0028,0011) are attributes in:
Table C.8-106 MR Spectroscopy Data Module <- not actually an image
Table C.7-11b Image Pixel Macro

Let's skip over MR spectroscopy for the moment, since it's not really
an image. (See Note in A.36.1.)

The Image Pixel Macro is referenced in:
C.7.6.1 General Image Module (Table C.7-9)
C.7.6.3 Image Pixel Module (Table C.7-11a)
C.8.13.1 Enhanced MR Image Module (Table C.8-79)
C.8.15.2 Enhanced CT Image Module (Table C.8-114)
C.8.19.2 Enhanced XA/XRF Image Module (Table C.8.19.2-1)

The first two are defined under C.7.5 Common Image IE Modules, so
they're Image IE for sure. For the last three, the Enhanced
Image Modules, look through PS 3.3 to see where each of
these are referenced in Image IOD Module Tables (in Annex A). In each
case, the module is shown as part of the Image IE.

So, Rows and Columns are attributes of the Image IE, not the Series
IE. So far, so good.

But, IODs are allowed to have modules that specialize attributes, and
the PET Series Module does just that. C.8.9.1.1.1 Specialization of
Image Plane Module and Image Pixel Module Attributes says
"For PET Series, the following Image Pixel Module attributes shall not
vary from Image to Image:
Photometric Interpretation (0028,0004)
Rows (0028,0010)
Columns (0028,0011)
Bits Allocated (0028,0100)
Bits Stored (0028,0101)
Pixel Representation (0028,0103)"

So, PET images must have consistent Rows and Columns throughout the
series. The others, not so much.

As for MR spectroscopy, it's not an image, but it does have Rows and
Columns. If you follow the same approach as above (left as an
exercise for the reader), you'll find that the MR Spectroscopy Module
is part of the MR Spectroscopy IE, which has the same relationship to
the Series IE as the Image IE does.

And as for Modality, that's a series-level attribute. But, now you
know how to find that out.

Hope this helps,
Dan Konigsbach
Manager of Framework Software Development
DR Systems, Inc.

I am extraordinarily grateful for your taking the time to lead me
through this point. I was unaware of the concept of Modules that
specialize attributes. I thank
you for your kind patience.

On the other hand, I have yet to see a CT scan or MRI with changing
numbers of rows and columns in a single series. I am glad this is
required for pet. I will think to put in a check for other image types
in the open source software display module I am writing.

Your kindness recalls for me the great helpfulness and patience that
nicely exists in members of the medical imaging community. I recall
the first time I met Gunter Zeilinger at RSNA many years ago and he
also was kind enough to explain to me aspects of the dicom standard.
Thank you again you have been very helpful!

Mitchell Laks