Myocardial infarction (MI) is one of the most common sudden onset heart diseases. Early diagnosis of MI is essential for management and treatment initiation. Electrocardiogram (ECG), as a noninvasive electrical recording of the heart behavior is one of the most reliable diagnostic tools for identifying patients with suspected MI. The QRS complex is the major feature of an ECG. There have been many researches for QRS detection algorithms. However, the current QRS detection algorithms have high false detections due to various types of noise or disturbances and sudden changes in the QRS complex.

We propose a novel QRS detection algorithm based on the use of simple pattern matching techniques in order to increase the accuracy of QRS detection. The algorithm aims to achieve better detection by grouping different ECG waveforms into 5 fundamental groups and then proceeding towards correction of detections based on this classification.

ECG had to be first filtered for high frequency noise and drift in order to be diagnostically useful. The filtered ECG is classified into standard and nonstandard groups using parabolic fitting. The QRS detection is performed on these groups. The algorithm proceeded by re-classifying the waveforms into 5 fundamental types of ECG. It then improved the detections using temporal correlation between successive ECG beats for further corrections. After all the appropriate corrections, identical waveform types on each lead were presented. The efficiency of the algorithm was also calculated from its true detection rate. QRS detection algorithm was tested using 20 MI patient data from the PTB diagnostic ECG database.

The algorithm resulted in a true detection rate of 98.9%. Our experiment showed that 199 leads among the 220 leads in 20 data sets were successfully classified into the five major groups. This proved to be a key step towards improving the accuracy of the algorithm as most of the waveforms belong to these major groups. As expected, our results confirmed that typical ECG waveforms are composed of successive ECG beats of similar patterns with little variation from one ECG beat to another.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error