A proposed methodology using dual-tree complex wavelet transform for feature extraction and a random forest classifier achieved an overall accuracy of 99.52% for classifying cardiac arrhythmias.
A random forest classifier using mixed temporal and frequency domain features achieved 99.52% accuracy in classifying cardiac arrhythmias from ECG signals.
In this paper, an essential and specific arrhythmias are identified and classified using electrocardiogram (ECG) signal. ECG provides information about the functionality of the heart. Abrupt changes in the shape of regular normal sinus rhythm (NSR) related to the cardiac beat is known as arrhythmia. Arrhythmia detection is a major challenge in medical field. A new technique is proposed in this work for the effective recognition of arrhythmias based on association for the advancement of medical instrumentation (AAMI) standard. In this work, the proposed methodology contains three crucial stages: (i) pre-processing, (ii) feature extraction, and (iii) classification. In the pre-processing step, noise is removed from the recorded ECG signal, which is effected from the various types of noises like baseline wonder, artifact, and muscle noises while recording ECG signal. Temporal (FS1) and frequency domain (FS2) features are extracted from the pre-processed ECG signal using dual-tree complex wavelet transform (DTCWT) in the feature extraction stage. FS1 is appended with FS2 (mixed feature set) and applied as an input to the random forest classifier for automatic recognition of cardiac arrhythmia beats in the last stage of the proposed methodology. The proposed work can classify arrhythmias with an overall accuracy of 99.52%.
Prakash et al. (Sun,) conducted a other in Cardiac Arrhythmia. Random forest classifier using mixed features (temporal and frequency domain via DTCWT) was evaluated on Overall accuracy of arrhythmia classification. A proposed methodology using dual-tree complex wavelet transform for feature extraction and a random forest classifier achieved an overall accuracy of 99.52% for classifying cardiac arrhythmias.