Phase-Lock Averaging for Noise Removal in PPG Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Physiological signals, such as PPG signals, are plagued by noise from various sources, including random noise from electronics and unpredictable physiological processes, which traditional Fourier methods struggle to effectively remove without introducing artifacts.
Innovation Solution
Phase-Lock Averaging (PLA) and Wavelength Standardization methods are employed to preprocess physiological signals, with PLA reducing noise by making assumptions about the pulsatile nature of the signal and Wavelength Standardization ensuring compatibility across different devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional Fourier methods are used to remove noise from physiological signals, then frequency-based filtering is achieved, but artifacts are introduced and noise removal is ineffective
Solution Approach 1:
The patent changes the fundamental parameter used for noise removal from frequency domain (Fourier) to time-domain phase-locking. By aligning signals based on physiological event timing (R-peaks, pulse onsets) rather than frequency components, the method achieves effective noise removal without introducing the artifacts characteristic of Fourier-based filtering approaches.
Solution Approach 2:
The patent replaces the mathematical transformation mechanism of Fourier transforms with a direct time-domain alignment and averaging mechanism. Instead of transforming signals to frequency domain and back, the system directly processes signals in time domain by aligning them to physiological events and averaging, substituting the Fourier transformation mechanism with a phase-locking mechanism.
2Measurement precision
If Phase-Lock Averaging is used to reduce noise in pulsatile signals, then signal quality is significantly enhanced, but the method assumes constant frequency and regular shape which limits applicability to irregular signals
Solution Approach 1:
The patent introduces dynamic adaptation mechanisms that allow the phase-locking method to handle variable frequency and irregular shape signals. By using detected physiological events (R-peaks, pulse onsets) as dynamic alignment points rather than assuming fixed periodicity, the system adapts to changing signal characteristics while maintaining noise reduction effectiveness.
Solution Approach 2:
The patent segments the continuous physiological signal into individual cardiac cycles or pulse waves based on detected events (R-peaks, pulse onsets). This segmentation allows each cycle to be processed independently and aligned to its specific timing, enabling the method to handle irregular frequencies and shapes while maintaining the benefits of phase-lock averaging.
3Quantity of substance
If measurements from different devices are integrated, then dataset size increases for model training, but wavelength variations between devices create incompatibility
Solution Approach 1:
The patent transforms the wavelength parameter from a device-specific fixed value to a dynamically calibrated value. By performing wavelength calibration for each device against a reference standard and applying transformation matrices, the system changes the wavelength parameters to a common reference frame, enabling compatible integration of data from multiple devices with different spectral characteristics.
Solution Approach 2:
The patent introduces a reference wavelength standard as an intermediary between different devices. Each device's measurements are transformed through a calibration process that uses the reference standard as a mediator, allowing measurements from devices with different wavelength characteristics to be converted into a common reference frame for compatible integration.
Data Source
AI summary
Methods and systems for removing noise from signal data obtained from a non-invasive blood monitor. In some implementations, the method may comprise receiving signal data from a non-invasive blood monitor and detrending at least a portion of the signal data to create a detrended signal. The detrended signal may then be stacked into a matrix. A singular value decomposition of the matrix may be taken of the matrix, which may be used to estimate one or more features of the signal that may be used to reconstruct the signal in some cases.