Text

DELINEASI SINYAL ELEKTRODIAGRAM MULTI-LEAD MENGGUNAKAN METODE LONG SHORT-TERM MEMORY BERBASIS EKSTRAKSI FITUR CONVOLUTIONAL NEURAL NETWORK

---

Penilaian

---

The ECG signal is a time-series data with very varied features where these features are divided into P waves, QRS Complex, and T waves. ECG signal delineation is a process of identifying the interval and amplitude positions on the features of each ECG signal waveform. Currently, the delineation of the ECG signal has been mostly done manually. However, this method hasn’t been good enough at delineating a huge number of ECG signal recording data. Much variety of feature data being processed manually allows misinterpretation to occur. In addition, processing a huge amount of data manually also consumes a lot of time. Therefore, a computer-based automatic delineation system is needed to overcome these problems. One method that is widely used these days is deep learning, therefore we will use deep learning on this computer-based delineation system. This study uses the combination methods between Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) where CNN is the feature extractor and LSTM is the classifier. There are two types of LSTM used in this study which are unidirectional LSTM and bidirectional LSTM (BiLSTM). Delineation is carried out on five-wave classes with 16 models designed to be trained and tested with Lobachevsky University Database (LUDB) data. Each model is designed with the best combination parameters of the hidden layer, batch size, learning rate, and epoch. The result shows that the model that produces the best results is model 16. The best model is the CNN-BiLSTM model with 15 CNN hidden layers and 2 BiLSTM hidden layers. This model is tested with parameters batch size 8, learning rate 0.0001, and epochs 300. This model produces the best evaluation results with the recall of 96.26%, precision of 95.15%, specificity of 99.30%, the accuracy of 98.94%, and F1 score of 96.20%.

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2207000428	T65083	T650832022	Central Library (Referens)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T650832022

Publisher

Inderalaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer., 2022

Collation

xiv, 137 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

005.740 7

Content Type

-

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Informatika
Pusat Data

Specific Detail Info

-

Statement of Responsibility

SEPTA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment