LEARNING NEURAL NETWORKS FOR DETECTION AND CLASSIFCATION OF BIOMEDICAL SIGNAL SECTION (ECG)
IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING,
2007, Volume 7, Issue 1, Pages 97-110
In this paper back-propagation neural network is presented for pattern recognition of ECG wave analysis and diagnosis, where training is applied for some common heart disease. Linear Predictive Coding (LPC) is used as a proposed method to compress the data, which were extracted from electrocardiogram, ECG paper. LPC method is tested before using it in this work, where it has succeeded in verifying coding operation to the signals. This method is efficient to reduce the ANN size used in this work. Data used are obtained from all currently available ECG databases, which were previously collected from different fields, such as Internet sites, different hospitals and some publications related with this field. The ECG samples were processed and normalized to produce a set of data that was applied to LPC and then to Artificial Neural Network (ANN). The results obtained are compared with the classifications made by a Doctor, where these results proved an efficient diagnosis with good performance and accuracy. Simulation results are obtained using technical (MATLAB) package implemented on IBM PC.
- Article View: 74
- PDF Download: 44