Accurate classification of cardiovascular diseases, particularly arrhythmias, remains a critical challenge in medical diagnostics. This paper presents a novel approach that transforms multi-lead ECG signals into Gramian Angular Field (GAF) images, which capture the temporal correlations within the signals by encoding them into visual patterns. The GAF transformation effectively preserves the temporal dynamics and amplitude variations of the ECG signals, allowing for a more comprehensive analysis. These images are then utilized for disease classification using hybrid deep learning architecture, which excel at identifying complex patterns. The proposed method demonstrates a significant improvement in classification accuracy, outperforming traditional techniques. This approach offers a robust and reliable tool for enhanced detection of cardiovascular conditions, with a particular focus on arrhythmia.