Detection of Artifacts and Brain Responses Using Instantaneous Phase Statistics in Independent Components

Model-Driven Development Methodology Applied to Real-Time MEG Signal Preprocessing System Design

“…Components related to cardiac artifacts are automatically identified using cross-trial phase statistics (CTPS). CTPS is performed by testing the distribution of phase values against a uniform distribution across trials and for each point in time [6], [38]. For identification of cardiac components using CTPS, the phase values of all the ICs are firstly estimated using the Hilbert transform [39] and band pass filtered within the range of 10-20 Hz, which is the frequency band of the QRS complex 1 [6].…”

“…The phase signals cluster near the time point of the R-peak when superposition of the phase trials from cardiac activity exists; in systems without any dependence on cardiac activity, the phase values will form a approximate uniform distribution whithout phase clustering. Details of the CTPS can be found in [6], [38].…”

“…It quantifies the probability that two data sets are samples of the same distribution. For checking stimuli-related phase responses, the cross-trial cumulative phase distribution is compared with a cumulative uniform distribution [38]. The Kuiper test algorithm was realized in Verilog using the Xilinx IP core library to ease the implementation of blocks as memories or arithmetic operators.…”