BiPath-CNet: Optimizing R-Peak Detection Leveraging Bi-Pathway Cardio Network with Spectral Temporal Attention Mechanism

R-peak detection is critical for accurate cardiac signal analysis, and traditional models often struggle with varying signal complexities. This study proposes a Bi-pathway cardio network (BiPath-CNet), an innovative deep learning (DL) model augmented with a Spectral Temporal Attention Mechanism (STAM) to enhance R-peak detection in electrocardiogram (ECG) signals. The proposed BiPath-CNet leverages STAM to dynamically capture spectral and temporal dependencies in ECG data, ensuring more precise feature extraction and robust peak detection. Extensive experiments demonstrate that BiPath-CNet performed better than several conventional DL models offering significant improvements in detection accuracy, recall, and precision. Integrating frequency-aware routing and temporal gating in STAM empowers the model to highlight critical signal components minimizing detection errors. This architecture offers a promising advancement in ECG signal analysis, enhancing the application of DL practices in biomedical research.