Cardiac Arrhythmia Class Segmentation Using the Deep 2D CNN Technique

In this article, we gave a strong technique to arrhythmia class division on electrocardiograms (ECGs) that utilizes profound two-layered convolutional brain organizations (CNNs), which have of late exhibited uncommon execution in the space of item identification and example recognization. As input for the CNN model, each ECG beat was converted into a two-dimensional grayscale image. Numerous deep-learning techniques are used to optimize the proposed CNN semantic segmentation model. Convolutional neural network (CNN)--based deep learning techniques have been effective in resolving a variety of issues in medical imaging, including image segmentation. Additionally, we contrasted our suggested model with AlexNet and VGGNet, two well-known CNN models. The evaluation employed the MIT-BIH arrhythmia database. At the evaluation, 10-fold cross-validation was carried out with each ECG recording serving as test data to precisely validate our CNN model. Our experimental results have effectively supported the claim that high accuracy may be reached without manually preprocessing the ECG signals with noise filtering, feature extraction, or feature reduction. This was done using the converted ECG images and the suggested CNN model.