Scholarlink

Circulation of Blood relies significantly precise electrical initiation. Arrhythmias can arise from any disturbance in the excitatory wave’s regular pattern in the heart. Electrocardiograms (ECG) are a popular diagnostic tool due to their speed, affordability, and non-invasive nature. This article presents a method for classifying ECGs that makes use of Convolutional Neural Network (CNNs) and the Continuous Wavelet Transform. Using Continuous Wavelet Transform (CWT), ECG signals are broken down into different time–frequency parts, while CNN extracts feature from the resulting 2D-scalogram. To further enhance the model considering the RR interval for arrhythmia diagnosis, 4 RR interval parameters are taken out and connected to CNN parameters to provide an ECG that is completely integrated classifier layer. This proposed model examines the impact of CNN and CWT on ECG classification alongside RR intervals. After training Accuracy of 97.27% was achieved. Based on the proposed method further models can designed for efficient clinical applications.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Electrocardiogram Classification Using Convolutional Neural Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kashyap Kachhadiya,
  • Abdullah Hasmani,
  • Drashti Gorasiya,
  • Bansari Patel,
  • Krishna Patel,
  • Vaishali Vadhavana

摘要

Circulation of Blood relies significantly precise electrical initiation. Arrhythmias can arise from any disturbance in the excitatory wave’s regular pattern in the heart. Electrocardiograms (ECG) are a popular diagnostic tool due to their speed, affordability, and non-invasive nature. This article presents a method for classifying ECGs that makes use of Convolutional Neural Network (CNNs) and the Continuous Wavelet Transform. Using Continuous Wavelet Transform (CWT), ECG signals are broken down into different time–frequency parts, while CNN extracts feature from the resulting 2D-scalogram. To further enhance the model considering the RR interval for arrhythmia diagnosis, 4 RR interval parameters are taken out and connected to CNN parameters to provide an ECG that is completely integrated classifier layer. This proposed model examines the impact of CNN and CWT on ECG classification alongside RR intervals. After training Accuracy of 97.27% was achieved. Based on the proposed method further models can designed for efficient clinical applications.