Weasis and ultrasound images coming from Acuson machine

[Marie ds]

Hi,
We have installed Dcm4chee 2.17 on a Mac server 1.5 years ago and it is working perfectly. Recently, I installed the weasis web viewer on the server and it is working fine. I noticed that some
 of our ultrasound images have a green color when we open them in weasis (but are fine if I open them with Osirix). It does not seem to affect the movie clips but only the fixed images. And the strange thing is that it does not affect all of the fixed ultrasound images. Our ultrasound is an old Acuson Sequoia and all of the images are coming from this same machine.
Searching the web, I found a post about this problem. It says that it is related to the decompression of RLE format with dmc4chee toolkit 1.x and 2.x. The last post is from 2009. I did not find any solution.
Did anyone came across a solution? Should I install another version of dcm4chee?
Any help will be more than welcome.
Thanks
Marie

[ime]

Can you upload some of these images in DICOM format to see the problem?

[Marie ds]

Hi,

Enclosed is one of the case with which the fixed images are displayed green in Weasis but are fine if we open them directly with Osirix.

Thanks for your help.

Marie

[nicolas...@gmail.com]

if I remember correctly, the problem is related to the Photometric Interpretation DICOM field (YBR_FULL that is a special color model used rarely) and the Java JPEG codec. Anyway, I can display correctly your images at least on Windows and Linux (see screenshot). Is the images displayed in green on Mac?

In my image test about Photometric Interpretation, there is a similar image and another one (YBR_FULL_422) where the thumbnail (jpeg image) has a problem with the color channel.

[Marie ds]

Hi,
According to the old posts that I have seen on the subject, yes it is related to the Photometric Interpretation DICOM field (YBR_FULL).
I have tried to open the same study with Weasis on PC and MAC and the problem is exactly the same: green tinge. I attached a screen shot of the study opened on a PC.
Thanks for your help
Marie

[nicolas...@gmail.com]

A more recent version of Weasis will display the images correctly. If you are using the Osirix web server, you can download the latest weasis-portable.zip, unzip it and replace the files at OsiriX.app/Contents/Resources/weasis/ with these in weasis-portable/.

[nicolas...@gmail.com]

A more recent version of Weasis will display the images correctly. If you are using the Osirix web server, you can download the latest weasis-portable.zip, unzip it and replace the files at OsiriX.app/Contents/Resources/weasis/ with these in weasis-portable/.

[Marie ds]

Hi,
Thanks for your help. I installed the new version of weasis and it seems to have worked. The images that were green before now only have a green contour (see image attached).
Marie

[nicolas...@gmail.com]

That's better. The problem of the green line is a scaling issue that had sometimes a border line (in the case the line is green because the value 0 is displayed in greed with this color model). Displaying the image at the actual pixel size (100%) should be OK. I will try to find a fix for this issue.

[nicolas...@gmail.com]

This issue has been fixed.

[Omar]

The issue of Photometric Interpretation DICOM field (YBR_FULL) (green tinge) came back in versions 2.0.0 and up to most resent. RLE compression Acuson ultrasound. 

[Nicolas Roduit]

Could you send me a sample

[Jamie Manley]

I can confirm that the problem does return beginning with 2.0.0.

With 1.2.9, the images are normal, but beginning with 2.0.0, the full size image is tinted green.

With versions 2.0.0 - 2.0.5, the thumbnails are correct, but the main viewer is tinted green.

Beginning with 2.0.6 and continuing into 2.0.7, the thumbnails and the images are tinted green.

I cannot send a sample image, but the problem is exactly as described in the sample images from Marie ds

[Nicolas Roduit]

The thumbnail state is not really meaningful because it could be built at the server side (right click on thumbnail and select "Rebuild Thumbnail" in Weasis)

I try with 2.0.7 and I don't see any issue with the set of images above.

First, check that your images have not be corrupted by another application (try to display in a non Java viewer). Otherwise, provide the procedure to reproduce the issue. 

[Jamie Manley]

I'm able to view them with RadiAnt, and with Weasis version 1.2.9. I haven't tried others yet.

I would upload them here, but I can't fully anonymize them (the patient name is embedded within the image from the Sequoia).

I did notice, after I posted this, the thumbnail is actually correct in 2.0.7, so that must have been a problem with 2.0.6 that was corrected by 2.0.7.

These images are being stored YBR_FULL not YBR_FULL_422 if that matters.

[Nicolas Roduit]

If the images are YBR_FULL and compressed, it is another problem that is related to java imageio library. This problem is fixed with the new embedded decoders in Weasis 2.5 (see what's new).

Here are some advantages of the new codecs:

• Compiled for Windows 32/64-bit, Linux x86 32/64-bit and Mac OS X 64-bit
  
• Fix https://java.net/jira/browse/JAI_IMAGEIO_CORE-126
  
• Fix https://java.net/jira/browse/JAI_IMAGEIO_CORE-189
  
• Fix artefact with native jpeg2000 decoder
  
• No need to patch jpeg-ls anymore
  
• Fix color model issue of jpeg-ls and jpeg
  
• Allow to get a signed data buffer (no need to patch afterwards)
  
• Allows to order the codecs by priority with a unique configuration for all the systems
  
• Supports multi-frame and multiple fragments at the same time

You can have Photometric Interpretation samples from https://dcm4che.atlassian.net/wiki/display/WEA/DICOM+Samples

• Run Weasis 2.5
• Run Weasis 2.0.7

[Jamie Manley]

Thanks for the links... I tried 2.5.0-rc1, and still have the problem. 

We did a test with John Doe so I could submit the image files to you (see attached ZIP file).

Also, see attached, that the images display fine in RadiAnt. I also used DICOM Converter from www.dicomapps.com to convert them to JPG.

[Nicolas Roduit]

Thanks for the sample. This is another issue related to the RLE compression. I've fixed the issue for next release.

[Jamie Manley]

Wonderful!!! Thank you very much!

I compiled your changes and it does work perfectly!!! I'll wait until the official release to deploy version 2.5.0 to production and I'll use 1.2.9 in the meantime.

Thank you again for your help!