Rekabentuk sistem pengecaman puncak isyarat elektrokardiogram menggunakan rangkaian hmlp berbilang untuk mendiagnosis kecacatan jantung
Loading...
Date
2006
Authors
Syed Hassan, Syed Sahal Hazli Alhady
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
ECG is a heart status analysis system used by cardiologist to interprets which is cheap,
effective, easy for implementation and safe to be used. The signal acquisition hardware
has been developed according to the standard for ECG signals acquisition. In this
research, suitable threshold for preprocessing of the signals was determined in two
stages (recognition of peaks aiid elimination of noisy peaks). The preprocessing stage
has successfully eliminated 92.71% of the noisy peaks from the ECG signals. MHMLP
network has been proposea in tRis research, to increase the performance in identifying
peaks (P, Q, R, S and T) of the ECG signals. By utilising the features selection
approach, the optimum peaks identification performance of the neural networks system
has been determined. From the implementation, the MSE of MHMLP network to
identify peaks of ECG signals has been recorded at -27.92d8. As a whole, MHMLP
network has achieved recognition performance of 83.78% and 89% during testing and
training phase respectively. In order to test the validity of the neural network approach
in identifying ECG peaks, and diagnosing the occurrence of heart disorder, an
implementation on pediatric patient with LVH was conducted. A total of 119 cases of
pediatric patient with L VH and 119 normal cases were used. Six features based on
parameter achieved during peak recognition process of ECG signal (P, Q, R, S and T)
were used as inputs to the HMLP network. The results of the diagnosis analysis show
that false positive is 5.08% and false negative is 3.39% during the testing phase.
Sensitivity and specificity during the training phase are perfect (100%}, while 96.55%
and 95% are acquired during testing phase respectively. The network performance is
perfect (100%) during training phase, while the performance during testing phase is
95.76%.
Description
Keywords
Pengecaman puncak , Mendiagnosis kecacatan jantung