Pipeline for IMOD Tomogram Annotations For ML #6599
Replies: 4 comments 5 replies
-
Hello @braxtonowens and welcome! I'm going to ping @brisvag since he has experience with CryoET, but in the meantime, re 1/2, have you checked the bfio and napari-aicsimageio plugins? They claim they can already open .mod files: https://www.napari-hub.org/plugins?sort=recentlyUpdated&readerFileExtensions=*.mod&page=1. For mrc files there's a bunch of plugins also: https://www.napari-hub.org/plugins?sort=recentlyUpdated&readerFileExtensions=*.mrc&readerFileExtensions=*.mrcs&readerFileExtensions=*.mrci&page=1 As for 3, we absolutely are interested in solving this use case. I don't know exactly the workflow you envision but there's a chance it's already feasible or very close to feasible with a combination of napari-threedee, clipping planes, plane rendering, and various annotation plugins such as BIOMAG. At any rate, it's probably also something that would be interesting to chat about live at one of our community meetings — see our meeting schedule — the next one convenient for US Eastern is in two weeks. Anyway, I'm excited you're interested and I'm looking forward to collaborating with you on this! I think a useful thing would be to play with the tools linked above and figuring out the pain points in using napari currently, and then we can see where best to change things to remove those! |
Beta Was this translation helpful? Give feedback.
-
@braxtonowens Thanks for the post. Please feel free to poke me (and Utz) some more about this as well. I'll second @jni's suggestions that @brisvag and @alisterburt know a lot about the current capabilities of napari + napari plugins for cryoET. |
Beta Was this translation helpful? Give feedback.
-
Some of these things are possible/on the horizon with my plugin for napari (blik. I haven't had the time or occasion to implement a reader for The code for arbitrary resampling of a volume (such as extracting an oriented subvolume) is implemented in a general library in https://github.com/morphometrics/morphosamplers, which I'm using to resample arbitrary filaments or surfaces with If you feel like contributing, I'd be happy to pair up :) |
Beta Was this translation helpful? Give feedback.
-
Hey @braxtonowens ! The only relevant thing I don't see mentioned above is https://github.com/teamtomo/imodmodel as a simple single purpose package for reading data from imod model files in Python - hope it's useful for you! |
Beta Was this translation helpful? Give feedback.
-
Imod has been the software of choice for tomogram annotations for many years. The Slicer feature specifically has made it easy for users to find and annotate 3d reconstructions of tomograms. Data collection teams lead by Grant Jensen have been using this feature to process the largest tomogram datasets in the world.
Recently The Imaging Institute at the Chan Zuckerburg Initiative published these datasets to the CrypET Data Portal. Unfortunately ML researchers(Me) wanting to construct supervised models for locating microstructures within these tomograms lack the necessary labeled data.
I believe this problem can be solved by the Napari Group. The main pipeline that I believe needs to be created looks like this:
More than just this feature request, I want to discuss different paths for creating this pipeline. I would love to add to this and contribute in any way I can. I feel that I would best serve napari by generating ideas for what features to include for step 3.
Beta Was this translation helpful? Give feedback.
All reactions