Source code for the paper "Automated Detection of Peripheral Artery Disease From High Resolution Color Fundus Photographs"

by Mueller, S et al.

File structure

include/ - scirpts providing classes for the creation of models, datasets and general convenience functions
scripts/ - standalone scripts that are used to prepare the data or generate visualizations
run_k_fold.py - main script to run the training using cross-validation
search_hyperparameters.py - script to search for optimal hyperparameters across one fold in the dataset
multiple_instance_learning.py - script to run the training process for the MIL model on a training and validation set
prepare_data.py - script to convert the raw dataset into folds and generate label files for the training

All training scripts are configured using the *.toml files.

Example for executing the k-fold training

python run_k_fold.py --data <path_to_data> --epochs 20 -k 7 -s "MIL" --model <path_to_pretrained_model>

Name		Name	Last commit message	Last commit date
Latest commit History 58 Commits
include		include
scripts		scripts
.gitignore		.gitignore
.gitmodules		.gitmodules
README.md		README.md
analysis_visualize_results.py.ipynb		analysis_visualize_results.py.ipynb
config.toml.old		config.toml.old
config_hyperparameter_search.toml		config_hyperparameter_search.toml
config_mil.toml		config_mil.toml
config_segements.toml		config_segements.toml
config_transfer.toml		config_transfer.toml
multiple_instance_learning.py		multiple_instance_learning.py
prepare_data.py		prepare_data.py
requirements.txt		requirements.txt
run_k_fold.py		run_k_fold.py
search_hyperparameters.py		search_hyperparameters.py
setup.py		setup.py
transfer_learning.py		transfer_learning.py