Key result
An Artificial Neural Network using Wavelet Transform for feature extraction achieved performance accuracies of 99.7% for training, 99.2% for testing known data, and 96.2% for unknown data.
A wavelet-based feature extractor combined with an ANN classifier achieved high accuracy (>96%) in automatically detecting cardiac arrhythmias from ECG beats.
Hypothesis-generating for wavelet-ANN arrhythmia detection; leaves open prospective clinical validation before adoption.
The heart plays the most vital role of supplying nutrients and oxygen in any organism. Any abnormality in its function renders the body to many complications which may sometimes even lead to death. Hence, timely and early diagnosis of any abnormality is extremely important. Another requirement of the hour is the Automatic detection. Several techniques have been developed till date, but efficiency achieved so far leaves room for improvement. This paper also, presents a technique that aims at automatic detection of cardiac abnormality using an Artificial Neural Network. The detection is done on the basis of the wave shapes of different QRS complexes for different arrhythmias which are extracted from the ECG beats using Wavelet Transform. As the Daubechies wavelets are similar in shape to the QRS complex of the ECG, db4 has been used in the above context. The performance accuracies achieved for training, testing known data and unknown data have been found to be 99.7%, 99.2% and 96.2% respectively. The MIT-BIH database has been used for the present study and an altogether of seven different beats have been used for classification.
No takes yet. Share an insight, caveat, or question.
Begum et al. (2014) studied Cardiac abnormality / arrhythmias. Wavelet based feature extractor and ANN based classifier was evaluated on Performance accuracy for unknown data. An Artificial Neural Network using Wavelet Transform for feature extraction achieved performance accuracies of 99.7% for training, 99.2% for testing known data, and 96.2% for unknown data.