ECG Decomposition using Cascaded Spline Projection Residual Auto Encoders

Kaveh, Samiee and Kovács, Péter (2023) ECG Decomposition using Cascaded Spline Projection Residual Auto Encoders. In: Proceedings of the 50th Computing in Cardiology. IEEE, Atlanta (GA), pp. 1-4.

Text CinC2023-2271.pdf Restricted to Registered users only Download (317kB) | Request a copy

Abstract

Recent advancements in remote or handheld patient monitoring devices have led to the development of novel domain-specific AI architectures that enable more accurate and faster real-time ECG diagnosis. We present a data-driven framework for decomposition of ECG signals based on B-Spline Variable Projection Neural Networks (VPNN) and cascaded residual auto encoders (AE). We use VPNN with B-spline bases in regressor mode. Hence, the output of each VPNN layer is an estimation of the input. ECG segment is passed through a set of cascaded VPNN regressors, where the input of each VPNN layer is the residual of the ECG segment and the output of its preceding VPNN regressor. In such a topology, the output of each VPNNcanbeinterpreted as a component of the input representing specific frequency and morphological characteristics. The effectiveness of the decomposition framework was demonstrated in detection of ventricular tachycardia (VT) and ventricular flutter (VFL) ECG segments with a sensitivity rate and a false positive rate of 92% and 9.5%, respectively.

Actions (login required)

Edit Item