» BEATS DOING IT YOURSELF
Adaptive Filters
Applied to Heart ECG


Navigation

Home
Abstract
ECG Filtering
Linear
Introduction
Theory
Methods
Results
Future
Parsing
Heart Rate Variability
Conclusion
Team
Code

The Theory

     behind this method is that the heart rate has a certain bandwidth, which bandpass filter can isolate. After this, the signal has a distinct pattern in the time domain. A heart pulse has a characteristic downstroke and then upstroke, so running a match filter with an example of a good pulse will strength this aspect of detection. The key characteristic of a heart pulse is that it is a short duration high amplitude pulse which can be detected by convolving with a small width boxcar. Finally because the signal is supposed to be a periodic heart rate, and thus pulse, once a general idea of the frequency has been established, convolving with a comb filter can help extract beats from noisy signals. This method can be applied recursively, with each pass detecting the heart rate more accurately, and altering parameters accordingly.






Template © 2002 Devil Design