Detection of fetal breathing movements with joint time-frequency analysis of cardiotocogram records

Most biological signals do not conform to the requirement of stationarity. This fact is the main reason for inappropriate results of analysis of these signals in the frequency domain. On the other hand researchers are interested not only in whether some periodical components exist but when they occur and how long they last. A lot of different joint time-frequency algorithms which are available do obtain such information but most of them produce complex structure of cross-term interference. The amplitudes of particular rhythms existing in biological signals have not only wide range but they change in time. These phenomena cause the frequently useful time-dependent pattern to be completely destroyed and become very difficult for further interpretation. An appropriate method of signal preprocessing may help the researcher to enhance some features and to obtain valuable information. Two simple algorithms of fetal heart rate filtering are described. Both of them allow to refine the fetal breathing movements component from complex structure of the joint time-frequency spectrogram