The Arrhythmia Detection Algorithm Study Based On Multiple Templates Matching

ECG (Electrocardiogram) parameters detection is an important aspect for clinical diagnosis of heart disease. The key problems in ECG signal processing are how to increase the accuracy of ECG signal acquisition, reduce noise interference and increase efficiency of ECG signal parameters detection. ECG signal processing is studied in this paper in three aspects which are:ECG signal pre-processing, ECG signal detection and parallel processing ECG.On ECG signal pre-processing, Baseline wander elimination is often one of the first steps required in the processing of ECG. Baseline noise occurs due to respiratory signal and body movements. Respiratory signal wanders between0.15Hz and0.5Hz frequencies. One of the most common methods to remove baseline wander is high pass filtering. This research review is based on high pass filtering of ECG signal to remove this baseline wander while preserving the low frequency ECG clinical information.On ECG signal detection, The paper introduces an arrhythmia detection algorithm based on feature extraction and multiple template matching. The algorithm creates the template database from the sample database. In order to create the feature database, the algorithm extracts the features of the template database, creates the indexes associated with the corresponding template database at the same time. Then, use templates of the feature database and the Electrocardiography (ECG) to be inspected to execute the rough matching algorithm. If the cross-correlation of the rough matching is bigger than the predefined threshold, use the indexes to load the corresponding template to execute the fine matching algorithm. The experimental results, by using MIT-BIH database to verify, show the algorithm can raise the detection accuracy.On parallel processing ECG, The paper proposed two parallel algorithms for Arrhythmia Detection Algorithm. Parallel algorithms utilizing Correlation Analysis were implemented to compare individual heartbeats and form heartbeat templates. The processing is done on a Cluster of two computers using the standardized Message Passing Interface (MPI) libraries. In the paper, the results of two different parallel approaches are compared, and their performance is evaluated.