Detection of Cardiovascular Diseases in ECG Images using Deep Learning Techniques

Abstract

This paper presents an intelligent system for automated ECG image classification, aiming to support the early and accurate detection of cardiovascular diseases (CVDs), the leading global cause of death. The system distinguishes four cardiac conditions: normal rhythm, abnormal rhythm, acute myocardial infarction (MI), and prior MI. It employs a two-stage approach combining deep learning and classical machine learning. Convolutional Neural Network, SqueezeNet, AlexNet, Xception extract high-level features via transfer learning, which are then classified using traditional algorithms. A lightweight web application, built with Flask and SQLite, enables secure user access and ECG analysis. Results highlight the effectiveness of this hybrid framework in enhancing diagnostic accuracy and clinical support.

Introduction

1. Background and Motivation:
Cardiovascular diseases (CVDs) are the leading global cause of death. Traditional ECG interpretation is manual and error-prone, especially in high-demand healthcare environments. This study introduces an automated system using AI (deep learning + machine learning) to interpret ECG images for faster, more accurate cardiac diagnosis.

2. Objective:

The system aims to:

• Automatically analyze ECG images

• Identify heart abnormalities early

• Improve consistency, reduce delays

• Enable remote and real-time diagnostics via a secure web interface using transfer learning and minimal training data.

3. Literature Review Highlights:

• Dissanayake et al.: Showed feature selection (e.g., backward elimination) significantly improves model accuracy.

• Ozcan: Demonstrated hybrid shallow and deep feature extraction for chest X-rays, adaptable to ECG.

• Khan & Nannavecchia: Developed efficient, lightweight CNNs for cardiac condition detection in portable systems.

• Bharti et al.: Used a hybrid ML-DL approach, improving accuracy with Isolation Forest and normalization.

4. Proposed System Overview:

The proposed end-to-end platform features:

• Image upload

• Deep feature extraction

• ML-based classification

• Real-time web results
  It surpasses traditional/manual ECG reading by automating diagnosis and supporting multiple cardiac classes (Normal, MI, History of MI, Abnormal Heartbeat).

5. Methodology:

• ECG Image Preprocessing: Resize to 128×128, normalize pixel values, label encoding.

• End-to-End CNN Model: Uses architectures like SqueezeNet, trained with Adam optimizer and callbacks (EarlyStopping, ReduceLROnPlateau).

• Hybrid DL+ML Model: Deep features from CNNs are input into ML classifiers (SVM, RF, KNN, etc.).

• Performance Evaluation: Metrics include accuracy, precision, recall, and F1-score.

• Web Deployment: Flask-based interface with login, upload, and result display functionality.

6. Results:

• SqueezeNet achieved 100% accuracy, precision, recall, and F1-score, outperforming AlexNet, CNN, and Xception.

• Hybrid models (e.g., CNN + RF, AlexNet + DT) also showed excellent performance.

• SVM and Naïve Bayes underperformed across models.

• SqueezeNet chosen for deployment due to speed, accuracy, and low computational requirements.

• Final model packaged as a .h5 file and integrated into the Flask app for real-time use.

Conclusion

This research demonstrates the feasibility and effectiveness of deep learning-based ECG classification for accurate cardiac diagnosis. The proposed system delivers high accuracy, real-time performance, and easy deployment through a web interface, making it a practical tool for clinical and remote healthcare use. By combining precision with accessibility, this work contributes meaningfully to early cardiac disorder detection and paves the way for scalable, AI-assisted diagnostic solutions.

References

[1] K. Dissanayake and M. G. Md Johar, “Comparative study on heart disease prediction using feature selection techniques on classification algorithms,” Appl. Comput. Intell. Soft Comput., vol. 2021, 2021, Art. no. 5581806 [2] T. Ozcan, “A new composite approach for COVID-19 detection in X-ray images,” Appl. Soft Comput., vol. 111, 2021, Art. no. 107669. [3] H. Khan, M. Hussain, and M. K. Malik, “Cardiac disorder classification by electrocardiogram sensing using deep neural network,” Complexity, vol. 2021, 2021, Art. no. 5512243. [4] Nannavecchia, F. Girardi, P. R. Fina, M. Scalera, and G. Dimauro, “Personal heart health monitoring based on 1D convolutional neural network,” J. Imag., vol. 7, no. 2, 2021, Art. no. 26. [5] R. Bharti, A. Khamparia, M. Shabaz, G. Dhiman, S. Pande, and P. Singh, “Prediction of heart disease using a combination of machine learning and deep learning,” Comput. Intell. Neurosci., vol. 2021, 2021, Art. no. 8387680.

Copyright

Copyright © 2025 Nathi Marygrace, G. Narasimham. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.