Biosignals of the heart are used in clinical routine thousand times each day: A standard 12-lead electrocardiogram (ECG) is a very common non-invasive diagnostic tool to analyse heart diseases. This can be done for short times in a clinical environment, but also several hours using so called Holter ECGs during the daily life of the patient. Especially in the latter case, this huge amount of data is usually evaluated by computer programs.
Another typical type of cardiac measurements are invasive intracardiac recordings (usually consisting of several channels) can be applied to patients in a clinical environment often parallel to a 12-lead ECG measurement. All signals are interpreted by a physician during a few minutes and this interpretation determines the next steps during treatment. However, the number of signals to be looked at simultaneously is huge and only very experienced cardiologists can interpret them fast enough and in the correct manner.
The aim of this work will be the processing of heart signals of different types focussing on the questions and needs of cardiologists. Better signal representations, new parameters and innovative assisting diagnose techniques will be developed to support the medical procedures. The focus is set on ECG signal processing but also other types of signals can be analysed. Furthermore, questions from research should be answered by processing signals automatically. This is performed with mathematical methods from the fields of signal processing, information theory, statistics and machine learning.