Non-linear Volterra Filter for ECG Reconstruction from Endocardial Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for reconstructing surface electrocardiogram (ECG) signals from endocardial or epicardial electrogram (EGM) signals are limited by their linear assumptions, which lead to inaccuracies in irregular rhythms and noise, failing to accurately capture the morphology of ECG signals and lacking a quality assessment for the reconstruction process.
Innovation Solution
The use of non-linear Volterra filters that introduce quadratic and cubic terms and delays to model the bidomain model of electrical signal propagation, allowing for a more realistic reconstruction of ECG signals from EGM signals, along with a method to determine and validate the filter parameters using least squares minimization and recursive least squares algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If linear filtering methods are used to reconstruct ECG signals from EGM signals, then the reconstruction process is simple and computationally efficient, but the reconstruction accuracy deteriorates in irregular rhythms and noisy conditions
Solution Approach 1:
The patent transforms the linear filtering approach into a non-linear filtering approach by applying the bidomain model, which changes the mathematical parameters and relationships in the reconstruction process. This allows the system to capture complex non-linear relationships between EGM and ECG signals, improving reconstruction accuracy in irregular rhythms and noisy conditions while maintaining computational feasibility through efficient algorithm implementation.
2Measurement precision
If non-linear Volterra filters with quadratic and cubic terms are used to model bidomain propagation, then the reconstruction accuracy improves, but the computational complexity and algorithm complexity increase
Solution Approach 1:
The patent segments the complex non-linear reconstruction problem into manageable components by applying Volterra filters that separate linear, quadratic, and cubic terms. This segmentation allows the system to handle each non-linear order independently, making the computational process more tractable while still capturing the full non-linear behavior of bidomain signal propagation.
Solution Approach 2:
The patent introduces dynamic adaptation by using recursive least squares algorithms that can adjust filter parameters in real-time based on incoming signal characteristics. This dynamic approach allows the system to maintain high reconstruction accuracy across varying rhythm conditions (regular and irregular) while optimizing computational resources through adaptive parameter tuning rather than fixed complex calculations.
3Ease of manufacture
If linear filtering is used for ECG reconstruction, then the method is robust and easier to implement, but it fails to accurately capture ECG morphology in irregular rhythms
Solution Approach 1:
The patent changes the fundamental mathematical parameters from linear to non-linear relationships, specifically incorporating quadratic and cubic terms that model the bidomain propagation physics. This parameter transformation enables the system to accurately capture ECG morphology in irregular rhythms while maintaining implementation feasibility through structured algorithm design and efficient computation methods.
Data Source
AI summary
An active medical device using non-linear filtering for the reconstruction of a surface electrocardiogram (ECG) from an endocardial electrogram (EGM) is disclosed. The device for the reconstruction of the surface ECG includes a plurality of inputs, receiving a corresponding plurality of EGM signals from endocardial or epicardial electrogram (x1[n], x2[n]), each collected on a respective EGM derivation of a plurality of EGM derivations, and at least one output delivering a reconstructed surface ECG electrocardiogram signal (y[n]), related to an ECG derivation, and a non-linear digital filter (12′, 12′, 14) with a transfer function that determines the reconstructed ECG signal based on said plurality of input EGM signals. The non-linear digital filter includes a Volterra filter type (12, 12′, 12″) whose transfer function includes a linear term (h1) and at least one quadratic (h2) and/or cubic (h3) term(s).