Goals and scope for CellCanvas

[kevinyamauchi]

I am creating this issue to discuss our goals and scope for CellCanvas.

From my perspective, CellCanvas is a software library for building geometric models of intracellular architecture. These models can be used for quantifying and comparing morphology as well as for input into simulations.

The core library contains:

• geometric models for all key intracellular components. This includes meshes (e.g., for membranes), networks of splines (e.g., for filaments), and particles (both position and orientation).
• data models for defining collections of the geometric models
• algorithms for extracting the models from imaging data (e.g., segmentation, pixel embeddings)
• a graphical user interface for extracting and building the models from images

Non-goals:

• provide state-of-the-art performance for all computer vision tasks. I think we should provide solid implementations of some core tasks, but also make it easy for users to import results from other algorithms
• perform simulations. I think we should be able to export the geometries to simulation software easily, but I don't think we need to have any sort of simulation engine in the core.

[kevinyamauchi]

What do you think @kephale ?

[kephale]

I'll muse on it a bit, but this sounds like where we're going.