Do not filter by SOPClassUID

Google Healthcare API, used by Imaging Data Commons, does not allow
filtering by SOPClassUID. So we cannot use that in the search filter.
We should think of alternatives so that we can include only WSI results.

These changes were needed alongside [this wsidicom PR](imi-bigpicture/wsidicom#149)
in order to view an example dataset.

The following were used for testing:

url = 'https://proxy.imaging.datacommons.cancer.gov/current/viewer-only-no-downloads-see-tinyurl-dot-com-slash-3j3d9jyp/dicomWeb'
study_uid = '2.25.25644321580420796312527343668921514374'
series_uid = '1.3.6.1.4.1.5962.99.1.3205815762.381594633.1639588388306.2.0'

Signed-off-by: Patrick Avery <[email protected]>

sources/dicom/large_image_source_dicom/assetstore/dicomweb_assetstore_adapter.py

@@ -333,7 +333,9 @@ def importData(self, parent, parentType, params, progress, user, **kwargs):
  msg = f'Invalid parent type: {parentType}'
  raise RuntimeError(msg)
 
- from wsidicom.uid import WSI_SOP_CLASS_UID
+ # FIXME: Imaging Data Commons will not allow us to filter by SOPClassUID.
+ # Is there any other way to filter out non-WSI results?
+ # from wsidicom.uid import WSI_SOP_CLASS_UID
 
  limit = params.get('limit')
  search_filters = params.get('search_filters', {})
@@ -346,12 +348,9 @@ def importData(self, parent, parentType, params, progress, user, **kwargs):
  series_uid_key = dicom_key_to_tag('SeriesInstanceUID')
  instance_uid_key = dicom_key_to_tag('SOPInstanceUID')
 
- # We are only searching for WSI datasets. Ignore all others.
- # FIXME: is this actually working? For the SLIM server at
- # https://imagingdatacommons.github.io/slim/, none of the series
- # report a SOPClassUID, but we still get all results anyways.
  search_filters = {
- 'SOPClassUID': WSI_SOP_CLASS_UID,
+ # See FIXME above
+ # 'SOPClassUID': WSI_SOP_CLASS_UID,
  **search_filters,
  }
  fields = [

sources/dicom/large_image_source_dicom/dicomweb_utils.py

@@ -31,18 +31,25 @@ def get_first_wsi_volume_metadata(client, study_uid, series_uid):
  image_type_tag = dicom_key_to_tag('ImageType')
  instance_uid_tag = dicom_key_to_tag('SOPInstanceUID')
 
- search_filters = {
- 'SOPClassUID': WSI_SOP_CLASS_UID,
- }
+ # We can't include the SOPClassUID as a search filter because Imaging Data Commons
+ # produces an error if we do. Perform the filtering manually instead.
+ class_uid_tag = dicom_key_to_tag('SOPClassUID')
+
  fields = [
  image_type_tag,
  instance_uid_tag,
+ class_uid_tag,
  ]
  wsi_instances = client.search_for_instances(
- study_uid, series_uid, search_filters=search_filters, fields=fields)
+ study_uid, series_uid, fields=fields)
 
  volume_instance = None
  for instance in wsi_instances:
+ class_type = instance.get(class_uid_tag, {}).get('Value')
+ if not class_type or class_type[0] != WSI_SOP_CLASS_UID:
+ # Only look at WSI classes
+ continue
+
  image_type = instance.get(image_type_tag, {}).get('Value')
  # It would be nice if we could have a search filter for this, but
  # I didn't see one...