IMPLEMENTASI CONTINUOUS WAVELET TRANSFORM DAN ARSITEKTUR SENET-GATED RECURRENT UNIT PADA KLASIFIKASI BEAT SINYAL ELEKTROKARDIOGRAM

LESTARI, ANNISA AULIA and Suprihatin, Bambang and Irmeilyana, Irmeilyana (2025) IMPLEMENTASI CONTINUOUS WAVELET TRANSFORM DAN ARSITEKTUR SENET-GATED RECURRENT UNIT PADA KLASIFIKASI BEAT SINYAL ELEKTROKARDIOGRAM. Undergraduate thesis, Sriwijaya University.

Preview	Image RAMA_44201_08011182126004_Cover.jpg - Cover Image Available under License Creative Commons Public Domain Dedication. Download (92kB) | Preview
	Text RAMA_44201_08011182126004.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (23MB) | Request a copy
	Text RAMA_44201_08011182126004_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (3MB) | Request a copy
	Text RAMA_44201_08011182126004_0026017102_0017057403_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (6MB)
	Text RAMA_44201_08011182126004_0026017102_0017057403_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (3MB) | Request a copy
	Text RAMA_44201_08011182126004_0026017102_0017057403_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy
	Text RAMA_44201_08011182126004_0026017102_0017057403_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (11MB) | Request a copy
	Text RAMA_44201_08011182126004_0026017102_0017057403_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (422kB) | Request a copy
	Text RAMA_44201_08011182126004_0026017102_0017057403_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy

Abstract

Recorded electrocardogram (ECG) signals can show normal or abnormal heartbeat patterns. One method to detect heart abnormalities in ECG signals is to classify heartbeats into five categories, namely non-ectopic beat (N), supraventricular ectopic beats (S), ventricular ectopic beat (V), fusion beat (F), and unknown beat (Q). This study combines preprocessing using Continuous Wavelet Transform (CWT) method to remove noise and classification of ECG signals in MIT-BIH Arrhythmia Database using a combination of SENet and GRU architecture. In this model, SENet is placed in the initial block to strengthen important features and reduce irrelevant features through squeeze mechanism and excitation block (SE) block. GRU is placed after SENet in order to capture signal patterns in the time sequence of the features that have been optimized by SENet. The signal quality improvement method applied in this study resulted in an SNR value of 28.43 dB, which falls into the high category, indicating that the signal quality obtained is very good with low noise levels. The proposed model showed excellent performance, with an average accuracy, sensitivity, specificity, precision, and F1-score of more than 90%. The accuracy value of 99.33% shows that the model can classify the beat ECG signals into each class very well. The sensitivity value of 98.33% shows the model's good ability to recognize data belonging to the five classes. The specificity value of 99.58% indicates that the model can recognize data that is not part of the five classes. The precision value of 98.35% indicates that the model has good accuracy in predicting arrhythmia disease for each class. F1-score of 98.33% shows the balance between precision and sensitivity. In this study, the performance results per class in classes F, S, Q, and V are very good in all evaluation metrics, namely 98.24%. In class N, the sensitivity value is still relatively lower than the others, which is 95.06%. Although the results for class N are low compared to other classes, the sensitivity value obtained is still very good because it is more than 90%. Based on the evaluation results, the combination of SENet and GRU architecture can be categorized as excellent in classifying the ECG signal beat into five classes.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	ECG Signal Classification, Continuous Wavelet Transform, Signal-to-Noise Ratio, SENet, GRU
Subjects:	Q Science > QA Mathematics > QA8.9-QA10.3 Computer science. Artificial intelligence. Computational complexity. Data structures (Computer scienc. Mathematical Logic and Formal Languages
Divisions:	08-Faculty of Mathematics and Natural Science > 44201-Mathematics (S1)
Depositing User:	Annisa Aulia Lestari
Date Deposited:	21 Mar 2025 06:48
Last Modified:	21 Mar 2025 06:48
URI:	http://repository.unsri.ac.id/id/eprint/169713

Actions (login required)

View Item