This repository contains the development of the final project of the Biomedical Engineering degree, by Matías N. Sosa and Cristian E. Morilla. The idea is to make a comparative study of different modern Deep Learning techniques to predict the pre-ictal period, that is, the period before the convulsive period. This will be done using EEG data.
The organization of the repository is as follows:
├───📁 src/ │ ├───📁 demos/ │ │ ├───... │ ├───📁 notebooks/ │ │ └───... │ ├───📁 sz_utils/ │ │ └───... │ ├───📄 visualizer.py │ ├───📄 config.py │ └───📄 setup.py ├───📄 .gitignore ├───📄 README-sp.md ├───📄 README.md ├───📄 Makefile ├───📄 requirements.txt └───📄 LICENSE.txt
In the src folder, we have the source code of the project. In the notebooks folder, we have the different jupyter notebooks that we have used to explore data and to think some processes. In sz_utils the are importat python files. In the preprocess file, we have the code that we have used to preprocess the data. data_handler is the module we made and use to manage the CHB data. The Makefile is used to automate the execution of the different scripts. The config.py file contains the configuration of the project, such as the paths to the data, the paths to the models, etc. The requirements.txt file contains the dependencies of the project. Finally, the visualizer.py file contains the code to visualize the results of the experiments. In the demos folder, we have some scripts to show how to use the different modules of the project.
conda create -n seizure-prediction python=3.10.4conda activate seizure-predictionpip install -r requirements.txtpip install -e src(this will install the project as a package)- Download the data from https://physionet.org/content/chbmit/1.0.0/. Using the terminal:
wget -r -np -nH --cut-dirs=3 -R index.html* https://physionet.org/files/chbmit/1.0.0/. If you already downloaded the data, you can skip this step. - Edit the
src/config.pyfile to set the paths to the data. What you need to edit it is theCHB_FOLDER_DIRkey from the config dict.
The running will be done using the Makefile.
make OPTIONThe different options available for now are:
web_app: Run the web app.
🤓 Matías Nicolás Sosa
🤓 Cristian Ezequiel Morilla