Kinetic gas molecule optimisation neural network for classification of electrocardiogram signals to identify heart disorder
Electrocardiogram (ECG) is an important biomedical tool for the diagnosis of heart disorders. Recent studies have worked a lot on designing automatic diagnosis systems to help physicians. However, automatic study of ECG patterns and heart rate variability is difficult due to the large variation in t...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 出版: |
2012
|
| 主题: | |
| 标签: |
添加标签
没有标签, 成为第一个标记此记录!
|
| 总结: | Electrocardiogram (ECG) is an important biomedical tool for the diagnosis of heart disorders. Recent studies have worked a lot on designing automatic diagnosis systems to help physicians. However, automatic study of ECG patterns and heart rate variability is difficult due to the large variation in the morphologies of heart waveforms, not only of different patients or patient groups but also within the same patient. The main objective of this research work is the automatic normalisation and classification of ECG signals of heart disorders. Two intelligent approaches based on Self Organising Map (SOM) and Particle Swarm Optimisation Neural Network (PSONN) are proposed to find the cutoff frequency for high frequency noise removal in ECG signals. 100 lead II ECG signals were obtained from the Physiobank database, composed of five types of ECG signals, including Normal, Supraventricular tachycardia, Bundle branch block, Anterior myocardial infarction (Anterior MI), and Inferior myocardial infarction (Inferior MI). For each ECG signal, the PSONN and SOM-identified cutoff frequency are applied to a Finite Impulse Response(FIR) filter, and the resulting signal is evaluated against the original clean and conventionally filtered ECG signals. The results show that the proposed intelligent system successfully denoised the ECG signals more effectively than the conventional method. |
|---|