mist-medical / MIST

MIST: A simple, scalable, and end-to-end framework for 3D medical imaging segmentation.
Apache License 2.0
29 stars 11 forks source link

Documentation update #24

Open fuentesdt opened 4 months ago

fuentesdt commented 4 months ago

Is your feature request related to a problem? Please describe. Summer students need documentation on python environment install. Most are not familiar the PATH variable to know where they are installing and which python is being used.

Describe the solution you'd like Need full path documentation installing MIST with mamba

Final MIST_SOP Draft 2.docx

Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

Additional context INSTALL MAMBA We initially installed conda, so we had to migrate everything over to mamba using the command Conda install – base -override- channels -c conda-forge mamba

The method we used was installing mamba using a pre-existing conda package, but in the future, to install mamba without conda, we’d have to install the Miniforge distribution, which comes with both the conda-forge and mamba preconfigured.

For mamba install, we’d have to use the following commands: curl -L -O https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh bash Miniforge3-$(uname)-$(uname -m).sh

or

wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh bash Miniforge3-$(uname)-$(uname -m).sh

Since miniforge is a part of condaforge, we are unsure if we can install mamba without some form of conda. However, mamba describes itself as a drop-in replacement for conda which may mean that we could install mamba without conda(?) Still unsure about it.

USING MICROMAMBA

Micromamba is a lightweight version of mamba that can be installed without conda. To do this we’d have to either install directly from the website, or use command line with:

Curl -L | bash Micromamba shell init -s bash -p ~/micromamba Source ~/.bashrc

And then we can create the micromamba env with

Micromamba create -n python= Micromamba activate myenv Micromamba install <some packages, ex. mist-medical>

PATHS

/full/path/to/conda - /rsrch3/ip/cmsantos1/miniconda3/bin/conda

/full/path/to/python3 - /rsrch3/ip/cmsantos1/miniconda3/bin/python3

INSTALL MIST We had to make sure we closed the command terminal where we installed mamba from because we tried installing mist on the same terminal and it wasn’t working. We opened a new terminal and ran this code: To install MIST, Python3 pip -m install mist-medical

CONVERT DATASET After we installed MIST, we needed to convert the dataset to csv format so that we can run it mist_convert_dataset –format csv –train-csv --dest . mist_convert_dataset-format csv-train-csv--dest

EDIT DATASET We edited the dataset json file “dataset.json”, change the places where its null to match data.

RUN 5-FOLD CROSS VALIDATION Finally, run 5-fold training with mist_run_all and it automatically does a 5-fold cross validation for you.