A Hybrid Deep Learning Framework Incorporating GBC-Optimized Attention-Based CNN and Image Processing Techniques for Accurate ECG-Based Cardiac Affliction Detection | ||
| JES. Journal of Engineering Sciences | ||
| Articles in Press, Accepted Manuscript, Available Online from 13 September 2025 | ||
| Document Type: Research Paper | ||
| DOI: 10.21608/jesaun.2025.392608.1541 | ||
| Authors | ||
| Shamna B* 1; Maheswaran C. P2; Anitha A3 | ||
| 1Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Nagercoil, Tamil Nadu, India | ||
| 2Department of Artificial Intelligence and Data Science, Sri Krishna College of Technology, Coimbatore - 641 042, Tamilnadu, India. | ||
| 3Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Nagercoil, Tamil Nadu, India. | ||
| Abstract | ||
| The necessity for effective and precise diagnostic methods to identify cardiac abnormalities has been highlighted by the rising prevalence of cardiovascular diseases. Electrocardiography (ECG), a widely used modality for assessing cardiovascular health, capturing the heart's electrical activity. However, interpreting ECG signals is often challenging necessitating advanced methods for reliable analysis. Therefore, this research proposes a novel Deep Learning (DL) approach for detecting cardiac afflictions in ECG imagery by integrating metaheuristic optimization techniques. In the initial stage preprocessing is performed, where ECG images are resized and denoised using Adaptive Bilateral Filtering (ABF) to enhance image quality. Also, Spatial Fuzzy C-Means (SFCM) Clustering topology is then employed for segmentation process, allowing precise isolation of relevant ECG signal regions. For feature extraction, the Gray-Level Co-occurrence Matrix (GLCM) approach is utilized, capture texture features that are indicative of cardiac conditions. Finally, the classification stage is performed using a Genetic Bee Colony (GBC) algorithm optimized Attention-Based Convolutional Neural Network (CNN),which enables the system to accurately identify and classify various cardiac abnormalities. The system is executed in Python software, and the outcomes provide superior performance than conventional techniques in terms of Accuracy of (98.21%) and performance analysis. | ||
| Keywords | ||
| Adaptive bilateral filtering; cardiovascular diseases; ECG signals; Fuzzy C-Means Clustering; GBC-CNN | ||
|
Statistics Article View: 8 |
||
