search

SyneRBI / SIRF-Exercises

SIRF Training and demonstration material
http://www.ccpsynerbi.ac.uk
Apache License 2.0
18 stars 21 forks source link

New geometry notebooks #111

Closed DANAJK closed 3 years ago

DANAJK commented 3 years ago

New geometry notebooks. Need copyright adding and possibly modifying data path. See get_nifti_data notebook for Zenodo location of data

KrisThielemans commented 3 years ago

thanks @DANAJK ... illuminating...

I'm updating the download location. Will push that soon.

KrisThielemans commented 3 years ago

Question after looking at a_geometry. I don't think that in the first part you actually say which coordinate corresponds to L. For instance is IPP[0] L, or IPP[2]?

Obviously related is the question what @rijobro intended when reversing the order noted in https://github.com/SyneRBI/SIRF/issues/906#issuecomment-865640054

KrisThielemans commented 3 years ago

pushed. I'm trying to see how it works when reading with sirf.STIR. If that works ok, I'll push that as well and then merge.

KrisThielemans commented 3 years ago

works ok, except that it fails on the oblique slices. I've added a warning.

Ready to merge?

DANAJK commented 3 years ago

To do this, it needs to know the patient position. This is available in DICOM data, but unfortunately not for NIfTI. The DICOM IPP, IOP etc can be computed from NIfTI data as shown in one of these notebooks. Maybe better to say that STIR and some other reconstructions set dummy values, so BEWARE?

KrisThielemans commented 3 years ago

Obviously related is the question what @rijobro intended when reversing the order noted in SyneRBI/SIRF#906 (comment)

Wrongly attributing to @rijobro here. Sorry. Discussion at https://github.com/SyneRBI/SIRF/issues/906#issuecomment-865640054

KrisThielemans commented 3 years ago

@DANAJK would you be able to create a README for this directory? If you use links to names of notebooks, it makes it easy for people to open them from there (currently only on GitHub, but presumably soon when using JupyterLab).

Do any of the registration notebooks need changing? If so, want to create some issues and even tackle it?

I suggest to merge this PR already such that it ends up in future builds of docker etc.

KrisThielemans commented 3 years ago

@danajk, could you shed some light on https://github.com/SyneRBI/SIRF-Exercises/pull/111#issuecomment-865647282? With the example data that you give, it really isn't clear which is which (at least to me).

DANAJK commented 3 years ago

IPP[0] is L. I could perhaps include a screenshot from OsiriX that labels the directions so that it is more obvious which phantom is where?

I don't know what to do with registration notebook - it has multiple areas that need addressing (cubic example, mis-leading flipping, test niftis that differ in scale as well as translation and orientation).

KrisThielemans commented 3 years ago

IPP[0] is L. I could perhaps include a screenshot from OsiriX that labels the directions so that it is more obvious which phantom is where?

great. what about just stating that? Whatever is easiest.

KrisThielemans commented 3 years ago

I don't know what to do with registration notebook - it has multiple areas that need addressing (cubic example, mis-leading flipping, test niftis that differ in scale as well as translation and orientation).

hmm.

what about

That might do the trick? Obviously would be better to have non-cubic images etc, but there's little doubt that the code will work with them, so I wouldn't try to do that now!

KrisThielemans commented 3 years ago

@DANAJK I'll merge this now. You can always fine-tune. The main thing that is missing is a README that explains what each notebook does (one-liner).

thanks a lot