Seizure Detection Tutorials

This series of notebooks, which are currently under development, demonstrate the application of signal processing and machine learning classification to epileptic seizure detection.

Currently four open-source datasets are used:

1. The Epileptologie Database¹⁸

2. UPenn and Mayo Clinic's Seizure Detection Challenge²¹

3. CHB-MIT Scalp EEG Database¹⁹

4. NEDC TUH EEG Seizure corpus²⁶

Other databases exist but have their limitations:

• The European Epilepsy Database (epilepsy-database.eu)

  • Big, Well documented
  • €3000/6000 3 year licence

• IEEG (https://www.ieeg.org/)

  • Open-source
  • Hard to navigate

• LONDI (https://ida.loni.usc.edu)

  • Has updated projects
  • Permission for access

Topics covered

• Biosignal Feature Extraction
• Feature Pre-Processing
• Supervised Classification
• Model Evaluation
• Hyperparameter Tuning (Gridsearch, Random Search, Bayesian Optimization)
• Ensemble Learning
• Dimensionality Reduction
• Batch Learning
• Multilayer Perceptrons
• Convolutional Neural Networks
• Recurrent Neural Networks

I'd like to give a massive thank you to the open source community for all the hard work that is put into the Python language, interactive computing resourses, and packages used to make this project a reality. Also thank you to Hvass Laboratories (http://www.hvass-labs.org/) whos Tutorials on TensorFlow were the inspiration for creating my own tutorials.

Progress Update

• 29/05/2020: Had some time to clean up the Dimensionality Reduction notebook which still has some work left to do on it (most of these notebooks are still work in progress). My current main focus is finishing my PhD (plan to submit in August) and submitting the work to journals. After I can release more of the code for these notebooks (particularly the one on creating the datasets). I also need to find a job, just in case anyone knows of anything relevent/interesting :)
• 07/03/2020: Added bits to the notebooks that existed and changed the ordering of topics a bit. There are still gaps in the notebooks that I want to work on but hopefully I will be updating these more regularly from now on. The main way of getting the data for the tutorials is to download them from my Google Drive, but I will release the code for creating them soon once I have tidied it up.

Getting Started

Due to the size of these notebooks you may need to use nbviewer to view the notebooks. To do this all you need to do is copy the url for the notebook (e.g. https://github.com/Eldave93/Seizure-Detection-Tutorials/blob/master/01.%20Overview%20of%20Datasets.ipynb) into the URL bar on the nbviewer website.

Prerequisites

The easiest way of interacting with these notebooks is to use Google Colaboratory.

"Colaboratory allows you to use and share Jupyter notebooks with others without having to download, install, or run anything on your own computer other than a browser" (https://research.google.com/colaboratory/faq.html).

I recommend Google Colaboratory mostly because of the size of the RAM available and the access to GPU's via the cloud. When working on the later notebooks, which use the TensorFlow package, the notebook will be ran a tonne faster! You can open the notebook in Colaboratory by clicking on the "Open in Colab" button at the top of the notebook.

Another option is to use Binder. Binder is a open-source cloud deployment for Jupyter notebooks (see https://mybinder.readthedocs.io/en/latest/ for details). Although completely free it does have relatively limited computational resources, with a maximum of 2GB of RAM. This means some of the later notebooks which use larger datasets will likely not fit into memory.

If you want to interact with the notebooks locally on your machine then I advise you download Anaconda with Python 3 (See http://docs.anaconda.com/anaconda/install/) to get you started. Anaconda makes it easy to create a virtual environment in which to install and manage Python packages. It also provides an easy interface in which to launch the Jupyter Notebook interface (the Anaconda Navigator).

Installing Packages

The notebooks start with a method to automatically install the required packages and data if using Google Colab. If working on these locally I encourage you to use a virtual environment. All you need to do is create a new environment and launch the Jupyter Notebook application from that environment (See http://docs.anaconda.com/anaconda/navigator/ for more information).

If you want to install the packages manually from command prompt (or Anaconda prompt) then install the following packages below.

• matplotlib
• pandas
• numpy
• scipy
• scikit-learn
• umap-learn
• imblearn
• seaborn
• mlxtend
• mne
• PyWavelets
• Tensorflow