Klasifikasi Sinyal Phonocardiogram Menggunakan Discrete Wavelet Transform (DWT) Dan Multilayer Perceptron Backpropagation (MLP-BP)

Bagas, Dwi Arifany (2021) Klasifikasi Sinyal Phonocardiogram Menggunakan Discrete Wavelet Transform (DWT) Dan Multilayer Perceptron Backpropagation (MLP-BP). Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text Cover .pdf Download (339kB)
	Text Abstract.pdf Download (6kB)
	Text Abstrak.pdf Download (7kB)
	Text BAB 1 .pdf Download (125kB)
	Text BAB 2.pdf Download (525kB)
	Text BAB 3 .pdf Download (345kB)
	Text BAB 4 .pdf Restricted to Registered users only Download (306kB) | Request a copy
	Text BAB 5 .pdf Download (7kB)
	Text DAFTAR PUSTAKA.pdf Download (77kB)
	Text LAMPIRAN.pdf Restricted to Registered users only Download (186kB) | Request a copy

Abstract

The heart is a very vital organ, lack of attention to heart health can lead to very dangerous diseases. According to a survey conducted by the World Health Organization (WHO) 71% in 2018 of all deaths caused by cardiovascular disease. Generally, electrocardiogram (ECG) signals are used as a tool for medical personnel to diagnose heart disease. However, the use of an ECG signal is relatively expensive and its application is difficult, so the use of a Phonocardiogram (PCG) signal can be a solution because its application is easy and the cost is relatively cheap. The use of electronic PCG signals requires extraction methods to reduce sounds other than heartbeats and classification methods to distinguish normal and abnormal heart conditions. This study aims to design an extraction method and a classification method as a tool for diagnosing heart disease using PCG signals. The design of the extraction method uses Discrete Wavelet Transform (DWT) and the classification method uses the extension of Multilayer Percepton Backpropagation (MLP-BP). The design and testing uses a dataset which is secondary data from http://physionet.org/content/challenge-2016/1.0.0/. The dataset consists of 2141 heart sound recordings with details of 1958 normal conditions and 183 abnormal conditions. Evaluation of the design results using parameters accuracy, sensitivity, specificity to variations of Mother wavelet (sym2, db2, coif 1), hidden nodes 8, 16, 24 and k-fold cross validation 5-10. The design of normal and abnormal heart classification based on PCG signals with mother wavelet configuration using symlet 2 decomposition level 5 and the number of hidden nodes 24 is able to produce an accuracy level of 93.92%, sensitivity of 95.38% and specificity of 70.86% Keywords : Heart disease, PCG signal, ECG Signal, DWT, MLP-BP

Actions (login required)

View Item