Multi-View Electrocardiogram Hash Coding

The electrical activity of the heart is a highly coordinated and integrated process, with various components intricately interrelated. Viewing multiple leads as distinct perspectives may lead to an underappreciation of this holistic nature. Therefore, this paper proposes a multi-view ECG hash encoding method. Specifically, based on the primary view of the original signal, this method constructs the secondary view of the ECG signal using autocorrelation analysis. Additionally, it constructs the tertiary view by employing fast Fourier transform (FFT), filtering techniques, and inverse fast Fourier transform (IFFT). Subsequently, convolutional neural network (CNN) features are extracted from these three views. It proposes an efficient multi-view hash encoding fusion learning strategy to generate a unified ECG hash encoding, which facilitates subsequent classification and detection tasks. The experimental results indicate that the proposed method outperforms currently popular approaches in terms of effectiveness.