الفهرس 
Noise and Artifact in the ECGOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
Medical advances have greatly reduced the threat of disease for human beings. However, incidental accidents are not easy to prevent. The deaths caused by these incidental accidents are not easy to prevent without a proper emergent rescuing mechanism. Consequently, researchers have developed several medical care systems to improve emergency care quality.
Electrocardiography (ECG) signal analysis is considered one of the core components of an integrated medical care systems. ECG diagnosis is one of the most valuable diagnostic tools.
This thesis presents a proposed design for an integrated electronic medical care system. This system uses digital system processing techniques to analyze ECG signals. This methodology employs Highpass Least-Square Linear Phase Finite Impulse Response (FIR) filtering technique to remove the baseline wander noise embedded in the input ECG signal to the system or reduce the noise as much as possible. Discrete Wavelet Transform (DWT) was utilized as a feature extraction methodology to extract the reduced feature set from the input ECG signal.
The design uses back propagation neural network as a classifier to determine whether the input ECG signal represents normal or abnormal ECG signal, by considering this technique as cheaper, more-accurate, and faster than using a chain of memory and search-retrieval circuits.
The whole system is implemented on Xilinx 3AN-XC3S700AN Field Programming Gate Array (FPGA) board. Necessary simulations for the implemented system have been done, the design is compared with some other designs in the literature review, indicating that the implemented system has a good accuracy compared to other designs, achieving total accuracy of 97.8%, and achieving reduction in resources utilized on FPGA implementation.
An input signal analysis is done, by interpolating number of samples in the input signal by factor of 2, and by decimating number of samples in the input signal to half, then comparing
the signal classification accuracy in each case, and by comparing those accuracies with the original signal classification accuracy obtained from the original proposed system.