Tomosynthesis

[dpla...@gmail.com]

My Windows-based OpenREM installation is failing to cope with Siemens and Hologic mammographic tomosynthesis objects.

The Orthanc log contains a pair of entries like this:

E0920 11:03:57.312387 HttpOutput.cpp:72] This HTTP answer has not sent the proper number of bytes in its body

E0920 11:06:58.029309 RunnableWorkersPool.cpp:78] Not enough memory to handle some runnable object

The DICOM objects are left in Orthanc, occupying a great deal of file space. The individual files are between 800 and 2000 MB in size.

Has anyone configured their Windows OpenREM server / Orthanc installation to cope with mammographic tomosynthesis objects?

Thanks,

David

[dpla...@gmail.com]

Perhaps I should arrange for these mammography x-ray systems to send RDSRs to OpenREM. Does anyone know if these systems can create and send RDSRs?

• HOLOGIC Selenia Dimensions
• SIEMENS Mammomat Inspiration

Thanks,

David

[Ed McDonagh]

Hi David

Regarding Orthanc - how up to date is your install? I don't have tomosynthesis objects being sent to my OpenREM install (more on that later!), but I do to my QA image server and they work fine. It is on Linux though, naturally! I migrated from my 32-bit Conquest install to 64-bit Orthanc install in order to receive tomosynthesis objects greater than 2 GB in size.

The other possibility is that you are running out of RAM when importing large objects?

Before I had access to RDSR from our Hologic systems I configured the proprietary projection view object to be sent to OpenREM (this was before the Breast Projection Object too). That object was useless for QA, but had all the dosimetry information for the tomo exposures, and they're really small!

Since about version 1.8, Hologic has been able to do RDSR and so we have been using these instead. I think there may have been a time when you had to have a particular licence or something, but I think it is available as standard now (as it should be...). 

I was surprised to find that the SIemens Mammomat Inspiration doesn't appear to support creating RDSR - the Revelation does, but that is using a different software to the Inspiration.

Kind regards

Ed

--
You received this message because you are subscribed to the Google Groups "OpenREM" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openrem+u...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/openrem/afb5d146-546f-4802-92f0-07a4f4b90437n%40googlegroups.com.

[dpla...@gmail.com]

Hi Ed,

Thanks for the message.

My Orthanc installation is 32-bit, so I think that is the issue. I will look to installing a 64-bit version. I will also investigate having the Hologic systems configured to send RDSRs to PACS so OpenREM can retrieve them instead of the image objects.

Kind regards,

David

[Gerd Lutters]

Hello David and Ed

On our windows installations  OpenREM (ORTHANC) receives mammographie RDSR from Philips, Horlogic and Siemens.

As far as we could test Philips and Horlogic are fine, 

Siemens Revelation, Siemens Inspiration only for MAMMOGRAM

SIEMENS Insight_2D and Tomo_proj do not give  dose values into OpenREM. 

Report from Siemens Revelation in OpenREM

L_CC AGD = 1.451

R_CC AGD = 1.566

L_MLO AGD = 0.053

R_MLO AGD = 0.053

and 

accumulated left AGD = 3.814

accumulated right AGD = 3.778

both AGD = 0 

Report from Siemens Inspiration in OpenREM

INSIGHD_2D ESD=0.00  AD= blank

and

accumulated AGD = blank

We verified against the machine display. Mammogram is ok but tomosynthesis is wrong. Does soemone know how to fix this issue?

On othert DMS we see simelar behavior. The reported dose for tomosynthesis is only about 1/26 of the dose reported at the display. Those number are consistend with dose reportted in the RDSR file (27 times a dose is reported, last line the accumulated ). 

best regards

Gerd

[Ed McDonagh]

Hi Gerd

Is there any chance you can send me an example Siemens RDSR - direct to me rather than the group.

Do you have any QA studies? Then I could have a look at what's going on.

Kind regards

Ed

To view this discussion on the web, visit https://groups.google.com/d/msgid/openrem/14b009c8-f1bb-44b6-806e-beb4556ce3f8n%40googlegroups.com.

[Ed McDonagh]

Thanks for sending the Revelation RDSR through. 

The numbers extracted into OpenREM reflect what is in the RDSR. An accumulated dose is presented for each breast (with a positive value), and both (with a value of 0).

There is then a sequence for the tomosynthesis event, with a reference to each of 26 images but only one AGD value, which is clearly not the AGD for the full tomo view. It is not 1/26th of the accumulated dose (despite there being only one view for that breast). It is also too small I think to be the dose for each of the 26 exposures with the rest being the pre-exposure pulse.

It would be helpful to see both what you expect the values would be for that exposure, and it would also be interesting to see if an RDSR with two tomo views on one breast has any way to distinguish the accumulated dose between them.

It might be that the BTO or the BPO if they are created might have more clues as to what is going on (though I wouldn't want to have to rely on them for routine dose collection!)

Kind regards

Ed