Filtering to find all H&E images in IDC

Support
1

Hello, I have an interest in seeing a list of all the H&E images in IDC. Is this possible?

Here’s what I’ve tried. I’ve used the exploration tools at the portal (https://portal.imaging.datacommons.cancer.gov/explore/) and was able to narrow down by selecting “Slide Microscopy” from Modality and “Brightfield illumination” from Illumination Type. Is there any further capability to filter specifically for H&E stained images? I have also tried the idc-index python package but again it is not clear to me whether there are any means to filter to the H&E stain.

Thank you very much for any suggestions!

1 Like
2

You cannot do this in the portal, but it is definitely possible with idc-index.

Take a look at this tutorial notebook: IDC-Tutorials/notebooks/pathomics/slide_microscopy_metadata_search.ipynb at master · ImagingDataCommons/IDC-Tutorials · GitHub.

idc-index sm_index table staining_usingSubstance group of attributes is what you need:

  • staining_usingSubstance_CodeMeaning: human-readable description of the stain
  • staining_usingSubstance_code_designator_value_str: colon-separated coding scheme designator and code value (typically, those code tuples will be coming from SNOMED-CT).

For the H&E slides, you will see the following, which you can use for filtering:

  • staining_usingSubstance_CodeMeaning: “[water soluble eosin stain, hematoxylin stain]” (in this or reverse order)
  • staining_usingSubstance_code_designator_value_str: “[SCT:36879007, SCT:12710003]” (you can look up what those codes mean and where they fit in the ontology of SNOMED-CT using this URL template: http://snomed.info/id/<code>, as in http://snomed.info/id/12402003 and http://snomed.info/id/36879007.

Note that in the source DICOM metadata, eosin and hematoxylin stains are siblings and are captured separately. In sm_index, these are aggregated into array to simplify access.

I hope this answers your question?

Did you see the tutorial notebook I mentioned earlier? If not, I recommend you consider other pathomics-focused notebooks available in the IDC-Tutorials repository: IDC-Tutorials/notebooks/pathomics at master · ImagingDataCommons/IDC-Tutorials · GitHub.

1 Like
3

This is an absolutely terrific answer, @fedorov! It took me a while to get to this but your advice/instructions are very clear. Thank you :tada:

1 Like
4

@daskelly I am so grateful to you for your note! Preparing those responses does take thought and time, and it is extremely encouraging to have feedback from the users - whether it is positive or negative, doesn’t matter! - just to know whether it helped and how we can improve.

Please keep the questions coming! We are here to help you make the best use of the resources we are developing. Your feedback is extremely valuable to help prioritize our development efforts!