Burst-suppression is a classic pattern in ICU EEG. This repository re-implements using a very simple algorithm based on thresholding local variance, and described in Real-time segmentation of burst suppression patterns in critical care EEG monitoring, Westover et al., 2013. Compared to the orignal article, here we also reject segmented periods shorter than 0.2 seconds (they are usually artefactual).
pyedflib, matplotlib, numpy
You can use records with burst-suppression from the TUH Abnormal EEG Corpus, or use another EDF file with burst suppression patterns. Link of the record to use in the parameters section of main.py.
6-second extract of a record with short and frequent bursts:
20-second extract of a record with longer and more spaced bursts:
