Electrocardiograph signal recognition using wavelet transform based on optimized neural network

<span lang="EN-US">Due to the growing number of cardiac patients, an automatic detection that detects various heart abnormalities has been developed relieve and share physicians’ workload. Many depolarization ventricles complex waves (QRS) algorithms with multiple properties have recently presented; nevertheless, real-time implementations in low-cost systems remain a challenge due limited hardware resources. The proposed algorithm finds solution for delay processing by minimizing input vector’s dimension and, as result, classifier’s complexity. In this paper, wavelet transform is employed feature extraction. optimized neural network used classification 8-classes electrocardiogram (ECG) signal data taken from two ECG signals (ST-T MIT-BIH database). coefficients are artificial network’s training process using invasive weed optimization (IWO) algorithm. suggested system sensitivity over 70%, specificity 94%, positive predictive 65%, negative more than 93%, accuracy 80%. performance classifier improves when neurons hidden layer increased.</span>