New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Various DICOM updates #2442
Various DICOM updates #2442
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks limited to Philips and DWI, and the field name (already present in our dictionary) certainly checks out.
Having multiple DICOM fields write to the same member variable however makes me nervous. In similar circumstances for other types of data (e.g. slice timing comes to mind), what I've done is to have different DICOM fields write to different member variables, and then only at the point of constructing the series, interrogate which fields have been filled comprehensively by which sources, and set the precedence of which shall be utilised if more than one source is available. Would this be a safer approach here?
I had similar reservations, but convinced myself that this particular variable ( But thinking about it further, you might be right that in some contexts where the |
We've also got much bigger problems in that the DICOM test suite fails currently, on |
OK, on closer inspection, I think the issue that screwed up the DICOM testing for the DW encoding info was #2231. Regardless, I'll need to look into this closely to figure out whether it's OK simply to update the test data. Also need to consider the artificial isotropic-weighted images that Philips likes to add at the end of their series (#2172). So far, I really can't see how to reliably do this - but I'll discuss that on the relevant issue. |
This also adds an environment variable to preserve these volumes in the rare cases where that might be desirable.
14b128c
to
5df4c98
Compare
You'll note that I've added a solution to #2172 in that last commit (5df4c98) - see this comment for a full description of the thought process... |
Turns out that last change breaks a lot of the testing, since the DWI dataset we use for testing was affected... Just to clarify: current master:
this branch:
Guess we're going to have to update some of the test data as well... |
This fixes test failures for the DTI tests that failed following the bug fix in e105aec. It does so by forcing bvalue scaling to be performed in those cases to emulate the previous behaviour. This may not be the best solution in the long term.
Right, looks like this is pretty much good to go. Decision required as to whether to go with this PR (see description in 4223ae6) or #2450. I personally favour #2450, I expect everyone else will too... Good news is both approaches check out, which gives us confidence that the updated test data on #2450 is correct. Can I get someone to approve #2450 (or this one if that's the preferred option)? Quick recap of the changes here:
I realise these should probably all be merged as independent PRs, but this should be OK as a single PR if I edit the title to e.g. 'various DICOM updates', right...? 🙏 |
@jdtournier Close? |
👍 |
Simple fix for issue reported on the forum, which lead to incorrect sorting of dMRI data for some Philips datasets.