Detection of sleep spindles in NREM 2 sleep stages: Preliminary study benchmarking of algorithms

Detection and classification of critical neural events during sleep is a central problem in EEG signal processing. Sleep Spindles constitute the most known pattern and their density in the EEG signal are related to many cerebral functions as memory consolidation, sleep quality or psychiatric diseases. Unfortunately this biomarker is underutilized because human annotation and classification is time consuming and almost impossible to achieve out of the scope of research. There is a need to use a reliable automated approach in order to use this biomarker in clinic.al practice A lot of studies and algorithms already exist and are used to help in this classification, but it remains difficult to achieve a good detection performance, especially when the EEG signal quality is low. We present here a review of the main methods used for spindles patterns detection and we test those where an open-source algorithm is available, to compare precision, recall and the F1-score on our own annotated dataset.