Simple Networks for Medical Imaging (SNMI)

A customized PyTorch framework for medical image analysis.

Quick start

Windows

1. Install Conda (https://www.anaconda.com/products/distribution)
2. Create Conda environment

conda create -n pytorch-env python=3.9

3. Install PyTorch through Conda (https://pytorch.org/get-started/locally/)

conda activate pytorch-env
conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch

1. Install dependencies

cd [your_path]/framework_pytorch
pip install -r requirement.txt

Tensorboard

The framework use Tensorboard to log and visualize results. While the training starts, run tensorboard on your PC:

cd [your_path]/results
tensorboard --logdir=log

If succeed, you will see something like:

Serving TensorBoard on localhost; to expose to the network, use a proxy or pass --bind_all
TensorBoard 2.9.1 at http://localhost:6006/ (Press CTRL+C to quit)

Then open the link shown above (http://localhost:6006/) in your browser to see the log and results.