I am investigating the tools currently available to view and interactively segment 3D data (made from lots of 2D slices), specifically for material composites from CT data. (So data that typically have a lot of similar features/cells in each image, but often shifted or transformed in some small way. See https://www.ndt.net/article/ctc2018/papers/ICT2018_paper_id156.pdf for some examples of what materials might look like)
From what I can see there does not seem to be a good extension to easily segment this kind of data in slicer. (Though I could very well be wrong on this and would love to be educated on it if that is the case)
The best segmentation I have thus far found for this kind of job seems to be:
though that tool is more a proof of concept than anything else. (It only runs in 1 thread and the viewer is minimal with a few bugs.)
I am considering taking the above code, optimizing it and turning it into a slicer extension. But before I do anything of that sort I figured I should inquire what the slicer community thinks. Have I missed any tools?, Are there better alternatives I should investigate first?
I am new to slicer, but have spent a few days playing around with it now and have tested out TomoSAM as well as a few other extensions for segmentation. TomoSAM does not seem ideal for this kind of thing, since you need to segment each individual feature it seems - though it does do a nice job of segmenting the individual feature on a pixel level.
I work for a non-profit called Alexandra Institute and our goal is to educate and develop tools to help danish companies with hard/interesting problems usually by bridging the gap between academia and industry.
So in this particular case, the goal would be to develop an open source interactive segmentation module to make it easier for companies to segment their CT image data of materials. (The 2 example datasets I have for this task both have around 600-700 images each with a resolution of ~4k4k - 5k5k in grayscale, so the data is rather large)