Template for DICOM SR

[Stefania Paperini]

Hi,

I have a question about choosing the right template for creating a DICOM SR.

The SR will contain the results of the automatic ACR-TIRADS analysis:

For each thyroid lobe: the lobe diameter, the lobe volume and the data of the segmented nodules: diameter, volume and the ACR-TIRADS classification results.

Which are the effects on interoperability if I define a private template, instead of choosing the TID1500?

For example, if a bridge sofware will be used to make available the DICOM SR data to the radiology software PowerScribe, will it also work if I define a private template?

Thank you for your help!

Stefania

[Antonio Magni]

Hi,

you need to define "work". The DICOM SR would still be in a standard format, so any DICOM SR viewer will show all of the attributes properly, just showing them. So that will "work".

If, though, there is a software that needs to rely on the meaning of the attributes and values of the SR (i.e. do something with them, instead of just showing them dumbly) as defined in TID 1500,  those will not "work" anymore.

You need to see what is missing from TID1500, so that you can then come up with a gap analysis and possible integrate them into the DICOM standard via change proposal: your needs might likely be needed by other as well.

[Andrey Fedorov]

> you need to define "work". The DICOM SR would still be in a standard format, so any DICOM SR viewer will show all of the attributes properly, just showing them. So that will "work".

> If, though, there is a software that needs to rely on the meaning of the attributes and values of the SR (i.e. do something with them, instead of just showing them dumbly) as defined in TID 1500,  those will not "work" anymore.

I don't think it is fair to say that just what is listed above would work. It really depends on what exactly individual implementation supports.

It is definitely possible, and there are applications, which will do way more than just show the attributes in the viewer.

In the research world, 3D Slicer (https://slicer.org) would take standard TID1500 reports containing segmentation-derived measurements, and will automatically pull the referenced segmentation object and segmented image series, load those, and will populate a table showing the measurements linked with the segmentation. OHIF Viewer (https://ohif.org/), for at least some variations of the TID1500, will display the list of annotations and corresponding measurements, and will allow navigating through those. The bottom line is the standard and SR templates - as is - are quite capable already. And what the original poster described - using TID 1500 to capture "For each thyroid lobe: the lobe diameter, the lobe volume and the data of the segmented nodules: diameter, volume and the ACR-TIRADS classification results." - should already work both with 3D Slicer and OHIF Viewer. If you share a sample, I can help with testing interoperability with the aforementioned open-source platforms. If you need public data to create a sample annotation - you will find plenty at https://portal.imaging.datacommons.cancer.gov/.

Andrey

[Antonio Magni]

I fully agree with Andrey. I'm not sure i understand the "i don't think it is fair" part. Sounds like we are both saying the same thing: it really depends on what the consumer of the data will be.

[Stefania Paperini]

Hi Andrey and Antonio,

Thanks for your answers! Yes, the question was about interoperability, so about what it is possible to do with the DICOM SR, apart from showing the attributes in a Viewer.

Andrey, it is very interesting to know that in 3D Slicer and OHIF Viewer it is possible to test the link between the segmentation and the SR, I didn't know that. Our study data include the source image as Enhanced US Volume (1.2.840.10008.5.1.4.1.1.6.2), the segmentation object as Segmentation Storage (1.2.840.10008.5.1.4.1.1.66.4) and the report as Comprehensive SR (1.2.840.10008.5.1.4.1.1.88.33).
The volume and the segmentation are already linked, so it is possible to overlay the segmentation to the volume, but the SR doesn't yet have any link to the source image and the segmentation. It would be a good idea to add that! 

If you want, I can send you the data created with a phantom and automatically segmented and classified, they are only for testing purpose and they are free for sharing, but I work for a company and I don't know if it might be a problem for you that the data are generated by a commercial software.

Antonio, you are right to write that to extract meaning from the data, the data must be known to the receiving system. As far as the TI-RADS values are concerned, I am not aware that there are codes for the classification that are in the Standard, as there are for the BI-RADS: https://dicom.nema.org/medical/dicom/current/output/chtml/part16/sect_cid_6027.html (Assessment from BI-RADS). At least, I was not able to find them. Asking for integrating them in the DICOM Standard is a good idea!

Stefania

--
You received this message because you are subscribed to the Google Groups "DICOM Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dicomforum+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/dicomforum/76e99838-3a61-4771-a965-94d5d4640491n%40googlegroups.com.

[Andrey Fedorov]

I sent a reply earlier, but used "Reply to author" button by mistake, and was wondering why my reply is not showing up ....

> I'm not sure i understand the "i don't think it is fair" part. Sounds like we are both saying the same thing: it really depends on what the consumer of the data will be.

Antonio, I think I mis-interpreted your earlier message as suggesting one cannot expect more than SR attribute visualization with the current level of support in DICOM. I am sorry about that! I am glad we are in agreement.

> it is very interesting to know that in 3D Slicer and OHIF Viewer it is possible to test the link between the segmentation and the SR, I didn't know that.

You can accomplish this by instantiating Row 7 of TID1411 https://dicom.nema.org/medical/dicom/current/output/chtml/part16/chapter_A.html#sect_TID_1411

You can find examples of image + SEG + SR that demonstrates how this can be done in IDC (https://portal.imaging.datacommons.cancer.gov/explore/), as an example, see this study: https://viewer.imaging.datacommons.cancer.gov/v3/viewer/?StudyInstanceUIDs=1.2.840.113654.2.55.194322112142778652977844374310761259090 (double-click on SEG to display segmentations). OHIF cannot display SEG-derived measurements, but Slicer should.

To download all of the files corresponding to the study linked above, you can do the following (assuming you have python environment set up):

pip install --upgrade idc-index && idc download 1.2.840.113654.2.55.194322112142778652977844374310761259090

> If you want, I can send you the data created with a phantom and automatically segmented and classified, they are only for testing purpose and they are free for sharing, but I work for a company and I don't know if it might be a problem for you that the data are generated by a commercial software.

It is not a problem for me. If you send, I will try to look.

Hope this helps

Andrey

[Stefania Paperini]

Hi Andrey,

I downloaded the study you mentioned and in 3D Slicer I can see the segmented data overlaid to the volume and the segmentation results table. It is very nice!
Also using OHIF Viewer I can see the segmented data overlaid to the volume, even if I got an error when loading the SR Reports.
Unfortunately when I try to load my Enhanced US Volume and the corresponding Segmentation in 3D Slicer and in OHIF Viewer, I get no overlay. Maybe I do something wrong, or the Enhanced US Volume is not fully supported.

If you want to have a look, I send you the data in attachment. The data were created with a phantom, so the segmentation is not the best, but there is a thyroid lobe and a nodule segmented. The DICOM SR is the actual version without any links to the other 2 Dicom objects. I will send you the updated SR version as soon as it is ready. The new version will also have TIRADS code from https://radlex.org/RID/RID50503 to improve interoperability (instead of private codes).

Thanks!

Stefania

 test_17_SEG.dcm

 test_17_SR.dcm

 test_17_USVol.dcm

To view this discussion visit https://groups.google.com/d/msgid/dicomforum/df204e73-8ce5-49e7-9644-abe36571f7b4n%40googlegroups.com.

[Andrey Fedorov]

Stefania,

I looked, but unfortunately I don't have sufficient experience with Enhanced US to help you troubleshoot.

Most likely it is not working in OHIF/Slicer simply because that specific modality is either not supported in combination with SEG/SR, or was not sufficiently tested.

Andrey