|
Jeff,
I have tried to implement the workaround but it is not possible to call the dicomparse
function. I need to add the path of the directory where this file is stored, but it is not allowed to add paths that are in a 'private' directory. Is there another possibility to call this function? Thanks in advance.
Jeroen
Jeff Mather <jeff.mather@mathworks.com> wrote in message <46E80930.1040905@mathworks.com>...
> Ingunn wrote:
> > I use dicomlookup to extract only certain tags from images
> > in a dicomdir.
> > My problem is when a tag is not present in the dicom
> > header, in that case dicomlookup stops the routine and I
> > get an error message:
> >
> > ??? Reference to non-existent field 'ScanOptions'.
> >
> > Error in ==> Rdcm>hListboxDirCallback at 244
> > dcminfo.ScanMode = gdata.metadata.(dicomlookup
> > ('0018','0022')); can mode
> >
> > Is there an easy way to work around the problem?
> > I'd like to avoid dicominfo because I'm only interested in
> > a few tags (to use in the routine, and for display on the
> > GUI)
> >
> > Thanks
> > Ingunn
>
>
> Hi Ingunn,
>
> The DICOMLOOKUP function only gives you the name that corresponds to a
> (group,element) pair, and vice versa. It doesn't actually communicate
> with any files or parse any data. You will still need to use DICOMINFO
> to parse the file.
>
> If you have more time than I do to devote to this task, you can actually
> hack up a pretty good solution that parses the whole file but doesn't do
> the slower post-parsing step that turns *all* of the raw data into what
> you would see with DICOMINFO. It's not hard to make this solution, but
> there's just too much going on here to actually put a robust, more
> usable solution like it into the toolbox. So, good luck...
>
> The DICOM parser in toolbox/images/medformats/private is quite easy to
> call and produces a well-contained output:
>
> attrs = dicomparse(fileDetails.name, ...
> fileDetails.bytes, ...
> getMachineEndian, ...
> false, ...
> dicomdict('get_current'));
>
> Be sure to call it with the correct arguments or you can get a
> segmentation violation, which is bad. Replace "fileDetails.name" and
> "fileDetails.bytes" with the filename and the number of bytes in the
> file. Replace getMachineEndian with either 'L' or 'B' depending on
> whether you're running on a little-endian machine (Windows, Linux, Intel
> Mac) or big-endian machine (other kinds of UNIX hardware). The "false"
> value performs some optimizations; I would recommend leaving it as is.
>
> The result is a structure array with these fields:
>
> * Group - A double containing the attribute's group number
> * Element - A double containing the attribute's element number
> * VR - A two-letter code giving the DICOM datatype (or empty)
> * Name - More or less the result of dicomlookup(Group,Element)
> * IsLittleEndian - 1 if the attribute contains little-endian data
> * Data - Either the raw bytes of the attribute's data or another
> structure of attributes containing sequence data
>
> If you go this route, your task is to search "attrs" for the group and
> element you're looking for, which is rather easy using something like this:
>
> idx = find(([attrs.Group] == myGroup) & ([attrs.Element] == myElement));
>
> What you do with the values in "attrs(idx).Data" is your business and
> will probably depend on what you want to do with the data. Because Data
> will contain UINT8 values, you will likely need to transform them into
> other MATLAB datatypes that correspond to the actual data intent. Look
> to the functions CHAR, TYPECAST, and SWAPBYTES for help. The
> translation rules require a fair bit of knowledge about DICOM, but you
> can likely find the information you need for simple types in the
> convertRawAttr() function in DICOMINFO.
>
> Good luck!
>
> Jeff Mather
> Image Processing Group
> The MathWorks, Inc.
|
Feed for this Thread