Classification of stereo-EEG contacts in white matter versus gray matter using recorded activity
A description of the method can be found in the accompanying paper.
To use, each patient's data must be saved as a separate matlab struct with the following fields:
name: a string, giving the patient ID or name, i.e. "patient 03"
data: < n_timesteps, n_contacts > array, giving the recorded voltage on every contact at each timestep
ch_type: < 1, n_contacts > array corresponding to the columns of data, with each entry 0, 1, or -1 indicating whether each contact is gray matter (0), white matter (1), or other (-1). "other" includes contacts that are outside the brain, malfunctioning, etc
ch_name: < 1, n_contacts > cell array, giving the contact name (as a string or char array) for each contact (corresponding to the columns of data)
shank_elecs: < 1, n_shanks > cell array, where shank_elecs{i} is an array of contact indices for shank i. These are the columns of data that correspond to the contacts on a particular shank. THESE MUST BE ORDERED FROM SHALLOW TO DEEP; i.e. shank_elecs{i}(3) is shallower in the brain (closer to skull) than shank_elecs{i}(4).
Fs: the sampling rate used to record this patient's data (in Hz)
Features are extracted and saved by running extract_features_script.m
The classifier is run on a given patient using main_script.m