Surgery – Diagnostic testing – Cardiovascular
Reexamination Certificate
2001-03-12
2003-09-23
Schaetzle, Kennedy (Department: 3762)
Surgery
Diagnostic testing
Cardiovascular
C128S901000
Reexamination Certificate
active
06625484
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a signal evaluation method for detecting QRS complexes in electrocardiogram (ECG) signals.
2. Background Art
Regarding the background of the invention, it can be stated that the automatic analysis of ECG signals is playing an increasingly larger role in perfecting the functionality of cardiac pacemakers and defibrillators. Newer models of implantable cardiac devices of this type accordingly also offer the capability to perform an ECG analysis. The detection of QRS complexes and R spikes in ECG signals plays an extremely important role in this context. This significance results from the many and diverse applications for the information concerning the time of occurrence of the QRS complex, for example when examining the heart rate variability, in the classification and data compression, and as the base signal for secondary applications. QRS complexes and R spikes that are not detected at all or detected incorrectly pose problems with respect to the efficiency of the processing and analysis phases following the detection.
A wide overview of known signal evaluation methods for detecting QRS complexes in ECG signals can be found in the technical essay by Friesen et al. “A Comparison of the Noise Sensitivity on Nine QRS Detection Algorithms” in IEEE Transaction on Biomedical Engineering, Vol. 37, No. 1, January 1990, pages 85-98. The signal evaluation algorithms presented there are based throughout on an evaluation of the amplitude, the first derivation of the signal, as well as its second derivation. For the presented algorithms, the essay distinguishes between those that perform an analysis of the amplitude and the first derivation, those that analyze only the first derivation, and those that take into consideration the first and second derivation. To summarize briefly, all algorithms check whether the given signal parameter exceeds or falls short of any predetermined thresholds, after which, if such an event occurs, the occurrence of additional defined events is checked based on a predefined pattern, and if certain criteria are fulfilled, the conclusion is drawn that an QRS complex is present.
Another aspect in the signal evaluation for detecting QRS complexes needs to be taken into account when methods of this type are implemented in implanted cardiac devices. In view of the natural limitations of these devices regarding their energy supply and computing capacity, it is important that the detection of QRS complexes can be performed with the simplest possible algorithms with the fewest possible mathematical operations on the basis of whole numbers instead of real numbers.
Signal processing methods from the fields of linear and non-linear filtering, wavelet transformation, artificial neural networks, and genetic algorithms have also been applied in the QRS detection. With large signal-noise distances and non-pathological signals, i.e., when good signal conditions are present, these evaluation methods produce reliable results. When no such conditions were present, the efficiency of the evaluation processes could drop drastically, which, of course, is not acceptable with regard to the reliable operation of a pacemaker.
SUMMARY OF THE INVENTION
Based on the described problems, the invention has as its object to present a signal evaluation method for detecting QRS complexes in ECG signals that can be used with a comparatively low computing capacity and also with problematic signal conditions while producing reliable detection results.
This object is met with the process steps according to the invention as follows:
sampling of the signal and conversion to discrete signal values in chronological order,
determining the sign of each signal value,
continuous checking of the signs of consecutive signal values for the presence of a zero crossing between two consecutive signal values,
determining the number of zero crossings in a defined segment of the consecutive signal values, and
comparing the determined number of zero crossings to a defined threshold value, wherein an undershoot of the threshold value is significant for the presence of a QRS complex in the defined segment of the signal curve.
The core element of the inventive method is the application of a zero crossing count that is based on utilizing the morphology of the QRS complex. The QRS complex in the ECG signal is characterized by a relatively high-amplitude oscillation that markedly guides the signal curve away from the zero line of the electrocardiogram.
The frequency of this short oscillation lies within a range in which other signal components, such as the P and T waves, exert only minor influence and can be removed preferably by pre-filtering, e.g., high-pass or band-pass filtering. After suppression of these low-frequency signal components, signal fluctuations result around the zero line, due to higher-frequency noise, that dominate in the region where no QRS complex occurs. The QRS complex then appears in this signal context as a slow, high-amplitude oscillation of only short duration. The differentiation between a QRS complex and the other signal segments can thus be detected with a frequency measurement that can be described representatively, based on the discussed signal characteristics, by the number of zero crossings per defined evaluated segment. The zero crossing count produces a number that is nearly proportional to the given dominant frequency of the signal.
In lieu of pre-filtering the signal values to eliminate the P and T waves, the QRS complex may, in the inventive method, also be distinguished from the P and T waves by determining the duration or the moment of the significant absence of zero crossings within the ECG signal.
The method of detecting the QRS complex by counting zero crossings has proven robust with regard to noise interference and easy to implement with respect to the computing technology. In this respect it is particularly suitable for implementation in the real time analysis of ECG signal morphologies in cardiac pacemakers.
The previously mentioned high-pass filtering is performed preferably with a lower pass frequency of 18 Hz. In this manner the low-frequency components, such as the P and T waves, as well as a base line drift can be suppressed. Furthermore, the QRS complex thus becomes the signal component with the lowest frequency that dominates the signal during its occurrence.
To increase the signal-noise distance, provision may furthermore be made to square the signal values prior to checking them for zero crossings and prior to determining the number of zero crossings, while maintaining their signs. As a result, smaller signal values are weakened relative to larger signal values, which further improves the detectability of the QRS complex.
The same purpose is served by the preferred method characteristic of the addition of a high-frequency overlay signal b(n) to the high-pass filtered ECG signal that has been squared while maintaining its sign. With this measure the ECG signal is manipulated in such a way that a number of zero crossings is attained outside the QRS complex that is significantly easier to predict. With a properly chosen amplitude, in particular, the ECG signal may be processed such that the number of zero crossings outside the QRS complex is identical to the number of signal values in the respective evaluated segment. This means that a zero crossing takes place between each sampled value, unless a QRS signal complex is detected at that time. This effect is increased if the high pass is additionally replaced by a band pass, preferably with lower and upper pass frequencies of 18 Hz and 27 Hz, respectively. The value of the amplitude of the high-frequency overlay signal is preferably determined adaptively from a flowing determination of the average of the band-pass filtered and squared signal values over a defined averaging period.
In accordance with a further preferred embodiment of the inventive signal evaluation process, the threshold value of the number of zero crossings signifying a
Köhler Bert-Uwe
Orglmeister Reinhold
Biotronik Mess- und Therapiegeräte GmbH & Co. Ingenieurbüro Berl
Browdy and Neimark , P.L.L.C.
Droesch Kristen
Schaetzle Kennedy
LandOfFree
Signal evaluation method for detecting QRS complexes in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Signal evaluation method for detecting QRS complexes in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Signal evaluation method for detecting QRS complexes in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3037710