Ventricular Fibrillation Signal Extraction via Ensemble EMD
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in extracting ventricular fibrillation (VF) signals during uninterrupted Cardio Pulmonary Resuscitation (CPR) is exacerbated by mode mixing and mode splitting effects in Empirical Mode Decomposition (EMD), which complicates the analysis of non-interrupted ECG signals due to intermittency and noise, leading to inaccurate decomposition and leakage of signal components.
Innovation Solution
A method and system that apply conjugate shadowing functions to ECG signals, using Empirical Mode Decomposition (EMD) to generate Intrinsic Mode Functions (IMFs), calculate their sum in different frequency regions, and divide by the number of shadowing signals to correct for mode mixing and splitting, thereby isolating the VF signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Empirical Mode Decomposition (EMD) is applied to non-interrupted ECG signals during CPR, then the signal can be decomposed into Intrinsic Mode Functions (IMFs), but mode mixing and mode splitting effects occur causing inaccurate decomposition
Solution Approach 1:
The patent introduces an ensemble of uniformly spaced phase signals as an intermediary to resolve mode mixing and mode splitting. By adding multiple phase-shifted versions of the original signal and decomposing each through EMD, then averaging the results, the method creates a mediating computational framework that separates signal components more reliably than single-signal EMD
Solution Approach 2:
The patent changes the parameter of signal phase by generating multiple versions of the ECG signal with uniformly spaced phase shifts (e.g., 0°, 45°, 90°, 135°, 180°, 225°, 270°, 315°). This parameter transformation allows the ensemble to capture different aspects of the non-stationary signal, resolving mode mixing when the results are averaged
2Ease of operation
If conventional EMD is used on ECG signals with intermittency and noise, then decomposition is achieved, but extrema locations become highly non-uniformly distributed causing mode mixing
Solution Approach 1:
The patent applies periodic action by generating an ensemble of signals with uniformly spaced phase shifts. This periodic transformation of the input signal ensures that extrema are more uniformly distributed across the ensemble members, and averaging them produces a result with more uniform extrema distribution than single-signal EMD
Solution Approach 2:
The patent merges multiple EMD decompositions by averaging the IMFs obtained from each phase-shifted signal. This combining operation integrates the information from all ensemble members, producing a final decomposition that overcomes the non-uniform extrema distribution problem of individual EMD applications
3Quantity of substance
If EMD decomposes signals with multiple frequency components, then IMFs are generated, but tone leakage into other IMFs occurs causing mode splitting
Solution Approach 1:
The ensemble of phase-shifted signals acts as an intermediary that prevents tone leakage. By decomposing multiple phase-shifted versions and averaging, the method creates a computational mediator that localizes frequency components more accurately than single-signal EMD, reducing mode splitting effects
Data Source
AI summary
This invention discloses a system and a method for extracting VF signal in ECG recorded during uninterrupted CPR. The present invention provides a method for extracting a Ventricular fibrillation (VF) signal in Electrocardiography (ECG), comprising: receiving an ECG signal; adding a plurality of shadowing functions to the ECG signal, to obtain a plurality of modification signals; decomposing the plurality of modification signals by using an Empirical Mode Decomposition (EMD) method, to generate a plurality of Intrinsic Mode Functions (IMFs); calculating the sum of IMFs in different frequency regions based on time sequence, dividing by a number of the shadowing signal, to obtain a plurality of modification intrinsic mode functions; combining the plurality of modification IMFs with the same property, to obtain a shape function; modeling the shape functions to obtain a compression signal; and subtracting the compression signal from the ECG signal based on time sequence, to obtain the VF signal.