Skip to content

JHancox/BioImaging

Repository files navigation

BioImaging

Whole Slide Imaging with RAPIDS

Prerequisites: See on ngc.nvidia.com/catalog/containers/nvidia:rapidsai:rapidsai for details

  • Notebook_1.ipynb - the first notebook to run uses cuCim to load the patient_100_node_0.tif
  • Notebook_2.ipynb - the second notebook, which uses dask to load foreground tiles and run them through an autoencoder
  • Notebook_3.ipynb - final notebook to build and visualize a graph from the embeddings generated
  • docker_build - folder containing a docker container build script
    • Dockerfile - file used to build a local Docker container that has all the dependencies needed

Steps to run the pipeline:

Launch a terminal in the root of the repo folder

Copy the following files into the data subfolder

cd docker_build

docker build -f Docker_Build -t dli/gtc23:v1 .

You can then run the container that this builds after returning to the root folder:

cd ..

docker run –gpus all –rm -it –ipc=host -p 8808:8888 -v [absolute path to current folder]:notebooks dli/gtc23:v1

The host folder should appear as 'notebooks' in the container, which means you can load and save things easily from the container. create a symlink:

ln -s /notebooks notebooks

This should make the folder appear in Jupyter

Once this is done, you should be able to simply browse to the jupyter lab using localhost:8808/lab

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published