Lost search results with qido + date filtering

[mdg583]

When using the qido-rs query method, filtering first in the qido query and then filtering those results by date and modality on the client will obscure search results in certain circumstances.

For example, if a user wishes to search for studies between two dates, and if qido-rs will be used with a limit filter of say 20 (such as may be the case for a paging system), then first a qido-rs query will be made without the date filter, and the first 20 such results will be returned to the client. Then the client will remove from those any that don’t match the date filter, leaving some number of results less than 20.

Even without setting the limit filter on the qido-rs query, this is still a problem, because servers are allowed to set their own maximum for the number of returned results. (On my test server this is 1000). If the number of studies on the server is more than 1000, then when using any date filter, only the matching results of the first 1000 studies are able to be found. The others seem to just vanish from the search.

[mdg583]

Since commit 1a264f8, the date problem is mostly fixed since the filter is applied on the server as well. I think the modality filter may be an issue still. I don't know if it is possible to filter date and modality on the server with the dimse option? I think that would be faster for date searches when the server has a large number of images.

[swederik]

Hi Matt,

Could you check if this modality filtering is still an issue? A filter has been added for the ModalitiesInStudy tag for DIMSE and QIDO. Note that ModalitiesInStudy may not be supported by some PACS (e.g. Orthanc).

Thanks for the bug report and your patch!

Erik

[mdg583]

Hello,
Sorry I won't be able to test this, as I've lost access to my test machine and I'm going to not be located at this medical center for a while, and I haven't been able to get this working in a virtual machine.

I modified some things a while ago to do the date filter in the queries as well, rather than on the client. I'm not sure what else this affects? It made using only the date filter a lot faster when there are lots of studies.
I tried to copy this set of changes by hand to make this patch, and so it definitely could have bugs.
studyDatePatch.txt

Also parsed patientId as an int so that its sorting happened correctly.

[swederik]

No worries. I believe I applied most of the patch manually. I don't remember changing patientId though so I will have to look into that.

Thanks for your contribution!