P300 Spelling and EEG Compressed Sensed Based on a Universal Megadictionary
Monica Fira, Victor-Andrei Maiorescu · Frontiers in artificial intelligence and applications · 2016
In this work, we propose, discuss, analyze and compare the results for a new EEG signal compressed sensed method. The EEG signal is recorded during a P300 based spelling experiment. The decompressed EEG signal is analyzed both in terms of decompression errors and spelling accuracy. The proposed method is based on building a universal mega-dictionary. Patient specific and inter-patient dependencies are analyzed and it is noticed that the results are not dependent on the patient.