cellpose
- Contributor(s): Carsen Stringer (@carsen-stringer)
- References: Project Site, Github, Documentation, Publication
Purpose
Describe the problem you are trying to solve by packaging your tool as a napari plugin in 1-2 sentences. I should be able to read this alone and communicate the value to someone else.
Why does this matter to biologists?
Segmentation of cells is an important step in a variety of biological processing workflows. A single segmentation algorithm may not work for all data types and therefore an easy way for a biologist to combine pre-processing steps and try a variety of deep learning algorithms is essential. Napari is an easy-to-use single package for the biologist to filter and process their images.
Why does this matter to you and your professional goals and responsibilities?
Providing biologists’ accessibility to complex tools is one of my major professional responsibilities as an algorithm developer.
Goals
What would you like to accomplish during the Alfa Cohort collaboration?
- Run 2D & 3D pre-trained cellpose in napari
- Save 2D annotations in a way that they can be loaded through cellpose CLI for training
- “How did I run this?” Support serialization of parameters for provenance tracking.
Non-goals
Are there any possible goals that are explicitly out of scope for the Alfa Cohort?
- Training cellpose model in the napari GUI
- Batch mode
Scope
Key Flows
Show what the end-to-end experience will be for biologists.
A biologist will be able to
- Load their data (2D or 3D)
- Choose a model (pre-trained from cellpose team OR specify their own)
- Choose what channels to run on & other parameters
- Run cellpose on the current data
- View the results
- Modify / curate data in a flexible way (handles on cells & reshape if they don’t like the boundaries)
- Save results for further analysis
A savvy biologist will be able to
- Load training data in 2D
- Annotate cells
- Export/save the annotations
- Run cellpose CLI on annotated data
- View results
- Apply trained model as above
Plan
Milestones
Status: To Do 📝, In Progress 🏗, In Review 🔎, Done ✅
Target Date | Milestone | Description | Status |
---|---|---|---|
2020-04-28 | Demo day | Demo final plugin | 📝 |