/CNN_sleep_stage_scoring

Classifying different sleep phases using the 6 EEG channels used in PSG studies

Primary LanguagePythonMIT LicenseMIT

CNN_sleep_stage_scoring

Classifying different sleep phases using the 6 EEG channels used in PSG studies, using amplitude envelope in 30s epochs. This is done on each individual channel to assess how valid they are in classyfying different stages. Further work would include PCA (perhaps a non-linear method like SOM) to assess the explained variance for each channel.

Funcs.py: The script containing the functions to filter, and detect outliers (i.e. motion artefacts).

Class.py: The python class where the user can define a class for the CNN model to train on, in a one-versus-all manner.

The label file should be in the format where each epoch is labeled at the time flagging it's start, so from 60 minutes to 60 minutes 30 seconds past beyond the starting point of the experiment, something like this: 60 : Wake

The recorded EEG also are going to be .edf files, which is the case for most available datasets.