Deep Learning-Based ECG Signal Quality Determination with Optimized CNN Hyperparameters

Electrocardiogram (ECG) signal quality assessment (SQA) is crucial for accurate cardiac diagnosis and monitoring. This study proposes a deep learning-based approach using a convolutional neural network (CNN) with optimized hyperparameters to evaluate ECG signal quality. The preprocessing phase involves resampling, segmentation, filtering, and normalization to enhance signal clarity. A comprehensive labelled dataset, comprising clean and corrupted ECG segments, is created to train and evaluate the CNN model. Various activation functions, including ReLU, Leaky ReLU, and ELU, are tested to determine the optimal model configuration. Performance metrics, such as accuracy, sensitivity, and specificity, are used to assess the model’s reliability across multiple datasets, including PhysioNet/Computing in Cardiology Challenge 2011 (PCCC2011) and INCART. Experimental results demonstrate the model’s high classification accuracy and robustness in distinguishing clean from noisy ECG signals. This automated quality assessment system is suitable for integration into real-time monitoring devices and telemedicine platforms, improving diagnostic accuracy and reducing the need for manual review.