troubles to visualize DICOM data

maria · Post by **maria** » Fri Jan 13, 2006 4:59 pm

I've got DICOM images, but I cannot open them in BrainVISA. The subject of working with MRI data and programs to analyze them is totally new to me so I do not really know how to describe the data. There is one folder, which consists of 1093 DICOM images. Then there is a file describing those images (called DICOMDIR). On all files there are no file endings. I only know that they are DICOMs because I was told so. The 1093 files are belonging to different study groups.
Is there a way of converting them so I can sort out those images belonging to the same study group?
Do the images need to have a certain file ending (for example .dcm)?
The scanner that was used is called 'Philips Imaging DD 001'. Maybe that helps.
For any help I'm really thankful.

maria

denghien · Post by **denghien** » Fri Jan 13, 2006 6:05 pm

Hi Maria,

You don't load yours data with Anatomist or BrainVISA because it isn't exactly the same problem.
Difference is that :
-> Brainvisa handles data and uses algo with them
-> Anatomist can load data to view them and manipulate things like referentiel, make fusion ....

I think your pb is you can't load data in Anatomist because when you load dicom data you must specify file type : 'all files' or 'DICOM' (it's the same) in the browser window.

Isa

maria · Post by **maria** » Mon Jan 16, 2006 11:45 am

I did specify the file type before I try to visualize the files and then they are shown in the file browser. But when I try to open them the following error message appers:

A.I.M.S. assertion failed: <int>colums == volDim [0], file c:/msys/1.0/home/demo/aimsdata-stable/src/aimsdicalm/io/dicomheader.cc, line 594
Directory where the file is saved: unrecognized format

I am able to view the files in another program called ezDICOM.

Maria

denghien · Post by **denghien** » Mon Jan 16, 2006 11:56 am

Hi,

I think that we could exchange DICOM datas to check loading and display.
About error message, it is the first time I see it.

Isa

maria · Post by **maria** » Mon Jan 16, 2006 2:46 pm

I found the problem. The images had file names in a different order (not consecutive). I renamed them regarding to their image number that I found in the header. Then everything went well.

Nevertheless, thank you for the help.

Maria

denghien · Post by **denghien** » Mon Jan 16, 2006 2:55 pm

Hi,

Congretulations

Isa

maria · Post by **maria** » Mon Jan 16, 2006 4:18 pm

Unfortunately, the procedure of importing DICOM data worked well for one data set but I do have 2 and for the second there occured another problem.

When using data management - import - anatomy - Import T1 MRI I specify an input file with the folder icon. Here I browse to the folder that contains all DICOM images belonging to the same series and click on one of the images. After filling out output and run the import process only the image I selected is being imported but not the whole series.
I've done exactly the same procedure with another data set and it worked.
Any suggestions ???

Maria

Manik · Post by **Manik** » Mon Jan 16, 2006 5:04 pm

This also happened to me a few times.
There seems to be some bugs in DICOM importation, so I use other software to convert the "bad" dicom images to Analyze or GIS format.

Post by **Yann Cointepas** » Mon Jan 16, 2006 5:17 pm

Thank you Manik for talking about "bad" DICOM images because the problem probably comes from a "bad" DICOM reader. Our home made DICOM input/output layer has been done years ago and is not very robust to the diversity of DICOM images. A more robust implementation is under construction but we are only a few developpers and we also have a research activity. Therefore things are evolving slowly.

denghien · Post by **denghien** » Mon Jan 16, 2006 6:26 pm

You can try to convert data from DICOM to GIS before importation with Aiims command.
Something like :
AimsFileConvert -i my_first_dicom_image -o my_volume.ima

This suppose that all dicom files are in the same directory.
And then you can use my_volume.ima to mak ean importation.

Isa

maria · Post by **maria** » Tue Jan 17, 2006 3:07 pm

I converted the DICOM images to Analyze format but the same problem occurs. I found out that everytime I import the data only the last image belonging to one volume is imported.

Post by **Yann Cointepas** » Tue Jan 17, 2006 3:27 pm

If you used AimsFileConvert for the conversion, it is not surprising that you have the same problem because Anatomist and Aims commands uses the same shared library for reading images.

I recently encountered the same problem for reading files in Genesis format (that is closed to DICOM format). I went around the problem by creating one directory per 3D volume (one per time step) and moving all the slices in these directories. Then Aims can read each volume individually and therefore AimsTCat can be used to put everything together. I wrote a Python script that do the job automatically. I can post it if you know Python. But I would suggest to check if it exists other free software to easily convert these DICOM images.

maria · Post by **maria** » Wed Jan 18, 2006 10:05 am

I found a rather weired solution. I used the program ezDICOM to convert medical images to DICOM format.
The reason I think this is weired is because the images were already in DICOM format. But as there has been written on the forum, maybe it was a 'bad' DICOM.

Post by **riviere** » Wed Jan 18, 2006 10:14 am

So it works with the DICOM images converted by ezDicom ? Well, in this case it's OK...
There are many variants of the DICOM format which have quite different shapes: some have one file per slice (we can only read this shape for now), others have one big file, and there are huge variations in headers information. This is why some programs such as ours can read some dicom images but not all of them. So if you could find a program that successfully converts your data into a format that brainvisa can read, it's fine.
Denis

denghien · Post by **denghien** » Fri Jan 20, 2006 3:44 pm

Hi Maria,

About your questions :

Display of DICOM :
I saw your dicom data. Although I sent you a mail and that I said I displayed the same slice from each file, now I obtained an other result when I download data once again.
It is a little strange but when I download data today by ftp, files have .dcm extension and not the first time.
The conclusion is today, I can :
-> display all slices of dicom volume with Anatomist. Perhaps that first time there had a transfert problem by ftp or something like that.
-> therefore I can convert all .dcm into GIS format!

Using of anatomical pipeline :
With this kind of volume, pipeline won't work correctly because characteristics are not consistent with basic data.
To use anatomical pipeline in common way, voxel size must be around 1x1x1 without strange artefact and a correct contrast. To have an idea about this, you can download lesson1 on http://brainvisa.info/downloadpage.html

Hope that helps

Isa