Medical device having automated ECG feature extraction
A medical device having automated electrocardiogram (ECG) feature extraction is disclosed. The medical device includes input circuitry configured to receive an ECG signal. Processing circuitry coupled to the input circuitry is configured to identify at least one fiducial point of heartbeat signature of the ECG signal. The processing circuitry is further configured to perform substantially simultaneously both a discrete wavelet transform (DWT) and a curve length transform (CLT) to identify the at least one fiducial point.
1. A method for a medical device having automated electrocardiogram (ECG) feature extraction comprising:
receiving by way of input circuitry an ECG signal generated by heartbeats;
processing the ECG signal by substantially simultaneously performing in parallel both a discrete wavelet transform (DWT) process that generates DWT data associated with a T wave and a P wave of the ECG signal and a curve length transform (CLT) process that generates CLT data associated with a QRS complex of the ECG signal by way of processing circuitry coupled to the input circuitry;
detecting an R peak of the QRS complex from the CLT data by identifying an amplitude within the CLT data that is greater than a QRS threshold updated as a portion of a sum of a previous QRS threshold associated with a previous one of the heartbeats and an average of CLT data generated from the previous one of the heartbeats; and
detecting a T wave peak and a P wave peak of the ECG signal from the DWT data by identifying within a forward search window following the QRS complex an amplitude that is greater than a T threshold updated as an average of the ECG signal within the forward search window and by identifying within a backward search window before the QRS complex an amplitude that is greater than a P threshold updated as an average of the ECG signal within the backward search window.
2. The method for the medical device having automated ECG feature extraction of claim 1 further including adjusting locations of boundaries of the forward search window and the backward search window relative to the QRS complex and an RR interval associated with the previous heartbeat.
3. The method for the medical device having automated ECG feature extraction of claim 1 further including extracting parameters that include PQ interval variability, QP interval variability, RT interval variability, TR interval variability, PS interval variability, and SP interval variability automatically by way of feature extraction circuitry coupled to the processing circuitry.
4. The method for the medical device having automated ECG feature extraction of claim 1 further including filtering the ECG signal by way of the input circuitry to substantially reduce noise having frequencies lower and higher than at least one wave within the ECG signal that includes the QRS complex.
5. A medical device having automated electrocardiogram (ECG) feature extraction comprising:
input circuitry configured to receive an ECG signal generated from heartbeats; and
processing circuitry configured to:
execute a curve length transform (CLT) that generates CLT data associated with a QRS complex of the ECG signal;
execute a discrete wavelet transform (DWT) that generates DWT data associated with a T wave and a P wave of the ECG signal, wherein the CLT and DWT are executed in parallel at substantially the same time;
detect an R peak of the QRS complex from the CLT data by identifying an amplitude within the CLT data that is greater than a QRS threshold updated as a portion of a sum of a previous QRS threshold associated with a previous one of the heartbeats and an average of CLT data generated from the previous one of the heartbeats; and
detect a T wave peak and a P wave peak of the ECG signal from the DWT data by identifying within a forward search window following the QRS complex an amplitude that is greater than a T threshold updated as an average of the ECG signal within the forward search window and by identifying within a backward search window before the QRS complex an amplitude that is greater than a P threshold updated as an average of the ECG signal within the backward search window.
6. The medical device having automated ECG feature extraction of claim 5 wherein the processing circuitry is further configured to adjust locations of boundaries of the forward search window and the backward search window relative to the QRS complex and an RR interval associated with the previous heartbeat.
7. The medical device having automated ECG feature extraction of claim 5 further including feature extraction circuitry coupled to the processing circuitry to extract parameters that include PQ interval variability, QP interval variability, RT interval variability, TR interval variability, PS interval variability, and SP interval variability.
8. The medical device having automated ECG feature extraction of claim 7 further including machine learning circuitry coupled to the feature extraction circuitry and configured to decide if a ventricular arrhythmia event is occurring based upon parameters extracted from the ECG signal.
9. The medical device having automated ECG feature extraction of claim 5 wherein the input circuitry is configured to provided band-pass filtering to substantially reduce noise having frequencies lower and higher than at least one wave within the ECG signal that includes the QRS complex.
10. The medical device having automated ECG feature extraction of claim 5 wherein the input circuitry includes a low pass filter to reduce low frequency noise from within the ECG signal.
11. The medical device having automated ECG feature extraction of claim 5 wherein the input circuitry includes a high pass filter to reduce high frequency noise from within the ECG signal.