Add URL to histogram bar tooltip to link to filtered dataset

Issue #139 resolved
Ed McDonagh created an issue

Objective: To identify outlying studies to investigate

How: Form URL complete with search criteria to display dataset. Will probably need to add new filtering methods.

Comments (14)

  1. David Platten

    This is a great idea that I think would be really helpful: being able to click on one of the bars in the histogram plot and be taken to the list of procedures that it is covering would be very useful.

  2. David Platten

    I have implemented this, but it needs testing on a larger data set than the one that I have available to me at the moment.

  3. Ed McDonagh reporter

    I've tested it on my live database using a test webserver, with several hundred DX exams, and it works quite well.

    I haven't thought about why, but the export function doesn't seem to take account of the extra filters.

  4. David Platten

    I’ve just tested it on the live system with 1600+ studies: it works as expected. The export does only include the extra filters in the initial test that I’ve carried out. I’ll test in more detail later.

  5. David Platten

    There are some problems with the display of data when a user clicks on a bar in the CT and DX acquisition histograms. This click goes to a link something like this: openrem/ct/?acquisition_protocol=Thorax&acquisition_dlp_min=1000&acquisition_dlp_max=2000. The problem with this is that it returns scans that have a Thorax sequence that has a DLP in the correct range, but also returns studies that have a Thorax sequence out of this DLP range, provided that there is another sequence that falls within the specified DLP range.

  6. Ed McDonagh reporter

    I had a related issue when I was creating the RF export code and I wanted to create the individual protocol sheets one at a time rather than as is done with CT where each study is sprayed across multiple sheets study by study.

    The solution for me was to obtain all the study_instance_uid that were of interest (code here) and then starting again at the Irradiation_event_xray_data model (code here)

    It would have to be a new filter in mod_filters.py as it a different model. But maybe it wouldn't work, as I don't know how you'd pass the uids. Bother.

    I was thinking this might have been a way of taming the data quantities to enable my size/CTDIvol scatter plots again, but I'm not sure.

    Maybe it would work - use the appropriate CT event model, and pass the other filters as before, but in mod_filters.py you'd work backwards for the name component to filter for the higher components such as station name etc.

  7. David Platten

    I've fixed the problem mentioned above about CT histograms not working - see commit 2e10a43. The problem was that for the study histogram I had to access a different array element to obtain the study names (0 rather than 2 for the acquisitions).

  8. Ed McDonagh reporter

    I had just tried this on a CT dataset and was going to report it wasn't working. But then I saw you had another commit: git pull, runserver - much better!

    However, whilst it now seems to filter CT correctly on DLP and acquisition protocol as per the graph, it doesn't carry through any of the other filters such as date or station name etc.

  9. David Platten

    The DX acquisition histogram list isn’t working on my live server – it works with my test system. I’m investigating that first, and then once it’s working I’ll add the other filters.

  10. David Platten

    Fixed DX acquisition histogram filter issue: missing DAP data caused a problem when converting to float. Fixed by excluding null DAP values before calculation of histogram. RE: issue 139.

    → <<cset 3478393076a4>>

  11. Log in to comment