Fig. 3. Algorithm applied for converting electrical signals into scalable touch activity. (A~F) continuous HKA for 24 h (left) and magnified 1 h activity highlighted with sky blue background (right). (A) Raw data file containing electrical noise. (B) Filtered data file showing clear touching signal after 1 Hz cutoff high pass filtering. (C) Absolutized data for further analysis. (D) Data reduction. Red dotted line shows threshold selected by Rosin's unimodal thresholding method. (E) Binary data (contact as 1, no contact as 0 determined by threshold) of continuous HKA for 24 h (left) and magnified 1 h activity highlighted with sky blue background (right). (F) Activity counts of continuous HKA for 24 h (left) and magnified 1 h activity highlighted with sky blue background (right). (G) Rosin's unimodal thresholding method used for auto-threshold detecting algorithm. (H) 20 Hz sampling rate simulation during 60 Hz Powerline Noise drift (59~61 Hz). Low frequency signals ranging from 0 to 1 Hz generated by drift of 60 Hz powerline noise simulation (bottom right). (I) Raw data trace (left top) and 1 Hz cutoff HPF filtered trace (left bottom). Power spectrum density plots of raw data (right top) and HPF filtered data (right top). Green colored traces are baseline without HKA signals and Red colored traces are baseline with HKA signals.
© Exp Neurobiol
