T. Oesterlein, T. Baas, H. Malberg, and O. Dössel. Multivariate AR model parameter estimation on time series extracted from the ECG of myocarditis patients. In Biomedizinische Technik / Biomedical Engineering (Proceedings BMT2011), vol. 56(1) , 2011
Biosignal analysis is aiming at analyzing physiological parameters for improved diagnosis and treatment. In this paper, the use of multivariate autoregressive models is proposed as a new method to analyze ECG data and to gain further information about the functionality of the heart. This application is demonstrated on myocarditis patients, where cure and diagnosis was observed. Timeseries of RR and QT intervals are analyzed by an autoregressive (AR) model, whose parameters are found dependent on the condition of the inflammation. The heart muscle inflammation is known as potentially lethal and an invasive biopsy still is the gold standard. Due to the variety of its symptoms, detailled non-invasive diagnosis is rather difficult and thus a highly challenging topic.
T. Baas, H. Köhler, H. Malberg, and O. Dössel. Automatic blood pressure segmentation algorithm for analysing morphology changes. In Biomedizinische Technik / Biomedical Engineering (Proceedings BMT2010), vol. 55(1) , pp. 168-171, 2010
Simultaneous recording of ECG, Atrial Blood Pressure (ABP) and respiration is possible and sometimes done to investigate cardiovascular and respiration coupling. But analysis most often concentrates on Heart Rate Variability (HRV) and Blood Pressure Variability (BPV). Although analysis of HRV and BPV has lead to important clinical information in the past, an investigation of the morphology of the time course of ECG and ABP could reveal additional diagnostic information.To analyse the morphology of the Blood Bressure (PB) wave a detection of characteristic points, outliers and boundaries is necessary. A wavelet based algorithm for blood pressure segmentation with outlier detection is presented in this paper. It is tested on 108 records with durations of 30 minutes each.