S. Süß. Determination of the information gain by estimating the instantaneous phase of biosignals. Institute of Biomedical Engineering, Karlsruhe Institute of Technology (KIT). Bachelorarbeit. 2018
Abstract:
When analyzing biosignals it is common practice to estimate the instantaneous phase (IP) in order to better follow signal propagation or to find phase singularities. However, the IP is often used despite the fact that the reliability of the methods for estimating the IP are not yet well known enough. Their effectiveness is therefore heavily debated. In this thesis, different methods for estimating the IP of biosignals are examined. For this purpose six methods for estimating the IP were applied to artificially generated biosignals mimicking intracardiac signals and those from the EIT. Subsequently, a signal processing study was performed in which noise was added and/or the signal morphology was varied. In the instance of intracardiac signals the effects of baseline drift were also studied. The resulting phase signals were then compared to the phase of the unaltered signals in order to test their robustness. Additionally, in the case of intracardiac signals the phase signals were compared to a linearly interpolated phase between Local Activation Times (LAT)s. The results show that the methods are reliable in the case of EIT signals. In comparison, the phase estimation of atrial signals was unreliable in many cases. As a result, two new methods based on the Non-Linear-Energy-Operator were developed that produce a far more reliable phase estimation of atrial signals. One of those methods is a peak-detection algorithm that interpolates the phase linearly between two peaks. It is shown that the phase estimation by peak detection yields results that are as good or better than the best preceding phase estimation method. This result was also confirmed when the methods were applied to clinical data.