oct-tissuemasking 0.0.5

A PyTorch based package for automated OCT tissue masking.

oct_tissuemasking

This package contains a basic 3D UNet and patching scripts to create training data.

1 Installation

1.1 Create a new mamba environment

Create a new mamba environment called oct_tissuemasking with python 3.9.

mamba create -n oct_tissuemasking python=3.9
mamba activate oct_tissuemasking

1.2 Install dependencies

We will need to install cyclopts (command line interface utility) and synthspline (for vasculature synthesis).

pip install git+https://github.com/BrianPugh/cyclopts.git#b862234
pip install git+https://github.com/balbasty/synthspline.git#f78ba23

1.3 Set cuda parameters

We need to identify and set our cuda version to make sure we install the right prebuilt wheel for cupy.

export CUDA_VERSION=<cuda-version>

1.4 Install oct_tissuemasking

This will take a while so we will set the timeout for a 20,000 seconds!

pip install oct_tissuemasking --default-timeout=20000