Pilot validation of a mimicking algorithm for detection of sleep spindles and K-complexes

Abstract

This paper presents a novel approach towards automated scoring of sleep EEG. Aiming to closely mimic manual scoring, a sleep specialist's approach to scoring is studied and mapped to equivalent logical steps using time- domain measures of the EEG. Unlike previous works, advanced signal processing tools are avoided as they result in derived measures which cannot be tuned easily at the user-end. Various modules have been developed to resemble the visual scoring methodology and then combined to produce an overall automated scoring algorithm. The algorithm was then tested on sample EEG data and its performance validated. © 2013 Springer-Verlag.