BIMCV code utils

This repository contains useful methods to handle medical images in Python, and implementations of Deep Learning models used by the Medical Imaging Databank of the Valencia Region team.

Modules

• Data: methods to handle datasets for developing DL models
• Dataloaders: PyTorch dataloader classes used by the training module
• Loss Functions: loss function definitions used by the training module
• Medical Imaging: methods to transform or extract relevant information from medical imaging formats (DICOM, NIFTI, ...)
• Metrics: metric definitions used by the training module
• Models: model definitions used by the training module
• MONAI: utils generated on top of the MONAI framework
• PyTorch: utils generated on top of the PyTorch framework
• Training: core module of the package, used to handle and log DL experiments. See usage instructions for more details

Installation

There are three options:

• Clone the repository and add the folder to the PYTHONPATH (recommended if you wish to contribute, or develop on top of the existing methods):

git clone https://github.com/BIMCV-CSUSP/BIMCV-AIKit.git
export PYTHONPATH="<PATH>/BIMCV-AIKit:$PYTHONPATH"

Replacing <PATH> with the proper value to generate a global path to the BIMCV-AIKit folder. The configuration of the PYTHONPATH may vary for your system.

• Clone the repository, and install from source as a pip package:

git clone https://github.com/BIMCV-CSUSP/BIMCV-AIKit.git
cd BIMCV-AIKit
pip install -e .

• Install as a standalone pip package using:

pip install git+https://github.com/BIMCV-CSUSP/BIMCV-AIKit.git#egg=bimcv_aikit

Dependencies

The current release depends on the following Python libraries:

• monai == 1.2.0
• numpy == 1.23.4
• pandas == 2.1.0
• prettytable == 3.9.0
• ptflops == 0.7
• pygad == 3.2.0
• tensorboard == 2.8.0
• torch == 1.12.1
• torchmetrics == 1.1.2
• torchvision == 0.13.1
• tqdm == 4.62.3

Install all dependencies using:

pip install git+https://github.com/BIMCV-CSUSP/BIMCV-AIKit.git#egg=bimcv_aikit[all]

License

This project is licensed under the MIT License.

Contact

• Jesús Alejandro Alzate-Grisales - Data Scientist
• Alejandro Mora-Rubio - Data Scientist