Scholarlink

<p>The heartbeats classification represents a very important tool in cardiology. Deep learning-based techniques for the analysis of ECG signals aid human experts in the appropriate diagnosis of cardiac diseases. This paper presents a two-dimensional deep learning technique for ECG heartbeats classification based on a convolutional neural network (CNN). We have considered five predominant kinds of ECG beats in MIT-BIH database, which are: Normal (N), Left bundle branch block (L), Right bundle branch block (R), Premature Ventricular Contraction (V), and Atrial Premature Contraction (A). The ECG recordings were first denoised and segmented into individual beats. Continuous wavelet transform was then applied to generate scalogram images from these beats, which were after used as inputs of the CNN neural network. The proposed system exhibits excellent performance, achieving a positive predictivity of 99.73%, a sensitivity of 98.02%, a precision of 98.68%, and an overall accuracy of 99.59%.</p>

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

ECG Heartbeats Classification Using Two-Dimensional Deep Learning Convolutional Neural Network

  • 期刊论文

  • 发表日期：

    2025年5月21日

查看全文
  • Yaaqoub Kahlessenane,
  • Fatiha Bouaziz,
  • Patrick Siarry

摘要

The heartbeats classification represents a very important tool in cardiology. Deep learning-based techniques for the analysis of ECG signals aid human experts in the appropriate diagnosis of cardiac diseases. This paper presents a two-dimensional deep learning technique for ECG heartbeats classification based on a convolutional neural network (CNN). We have considered five predominant kinds of ECG beats in MIT-BIH database, which are: Normal (N), Left bundle branch block (L), Right bundle branch block (R), Premature Ventricular Contraction (V), and Atrial Premature Contraction (A). The ECG recordings were first denoised and segmented into individual beats. Continuous wavelet transform was then applied to generate scalogram images from these beats, which were after used as inputs of the CNN neural network. The proposed system exhibits excellent performance, achieving a positive predictivity of 99.73%, a sensitivity of 98.02%, a precision of 98.68%, and an overall accuracy of 99.59%.