Modified Structural Intensity for Singularity Localization in Noisy Signals: Application to Coherent Averaging for Event-Synchronous ECG Interference Cancellation in Diaphragmatic EMG Signals

This paper addresses the problem of singularity location in a noisy signal. Landmark alignment performances of a modified structural intensity formulation are analyzed in the context of 'averaged template subtraction'. The chosen application is the event-synchronous ECG interference cancellation (ESC) in the diaphragmatic electromyographic signal. The structural intensity event-synchronous interference canceller (SIESC) algorithm based on the modified structural intensity formulation is compared with an existing ESC algorithm that requires an ECG reference signal for the coherent averaging process. Efficiency in interference cancellation is evaluated using both simulations and real electromyogram of the diaphragm data. All the results show that the alignment error of the SIESC induces an acceptable discrepancy between the two methods. One interest of this application is to demonstrate the ability of the structural intensity to provide a detector combining robustness and precision. A drawback of this approach is the calculation cost.