الفهرس | Only 14 pages are availabe for public view |
Abstract The electrocardiogram (ECG) is a record of the electrical activity of the heart measured at the body surface. The recording of the ECG waveform generally extends over a considerable time period, and in terms of reducing transmission time/storage requirements. It is clearly advantageous to apply data compression to ECG signals. A new algorithm for ECG signal compression is introduced. This algorithm IS based on the main features of direct method with new data selection and reconstruction method based on the advantage of using neural network as an interpolator. The algorithm is compared, using the same database and the same number of selected samples with the conventional transformation methods like discrete cosine and wavelet transforms. Comparison between the transformation methods and our work is based on two measuring values. These measuring values are the PRD and the diagnostability. This thesis consists of seven chapters given as follows: Chapter 1 is a general introduction concerning reasons for ECG signal compression, methods of ECG signal compression and reasons for using neural networks in our work. Chapter 2 is a summary of the fundamentals of ECG signals, heart diseases and modeling signal. Chapter 3 is a summary of the basis of ECG signal compression methods and measures of their fidelity. Chapter 4 is a description of the new peak picking algorithm which is based on linear interpolator with different compression levels and the results of applying this method on selected heart diseases. Chapter 5 is a neural network application based on backpropagation algorithm to interpolate the received ECG samples. Chapter 6 is a comparison between our two interpolators and the standard transformation methods (DCT, WT). Chapter 7 is a general conclusion of our research. |