Wavelet Transform Scale-Dependent Archetyping for Physiological Signal Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current signal processing methods for physiological signals, such as photoplethysmograph (PPG) signals, face challenges in extracting useful information due to noise and the need for knowledge of underlying signal periodicity, which limits their effectiveness in determining parameters like pulse rate and oxygen saturation.
Innovation Solution
The method employs a continuous wavelet transform using a wavelet function to generate an archetype transformed signal, which is computed using a weighted averaging scheme based on the natural periodicity of the wavelet, allowing for scale-dependent signal averaging without requiring knowledge of the signal's periodicity, and generates an archetype scalogram to derive physiological information like pulse rate and oxygen saturation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional signal processing methods are used to extract physiological information, then the processing is simpler, but the measurement precision and reliability are reduced due to noise interference and requirement for knowledge of signal periodicity
Solution Approach 1:
The patent applies segmentation by decomposing the physiological signal into multiple scale components through wavelet transform. Each scale represents different frequency bands of the signal, allowing selective analysis and processing of specific physiological features (e.g., pulse rate at certain scales, respiration at others) while filtering out noise at other scales. This multi-scale segmentation improves measurement precision without requiring prior knowledge of signal periodicity.
Solution Approach 2:
The patent transitions from time-domain analysis to time-scale analysis by introducing the wavelet transform domain. This adds a scale dimension to the signal representation, enabling visualization and analysis of physiological parameters across different time scales through the scalogram. This dimensional change allows better separation of signal components and noise, improving extraction accuracy.
2Adaptability or versatility
If signal processing requires knowledge of underlying signal periodicity, then the processing can be simplified, but the adaptability is reduced when dealing with varying physiological conditions
Solution Approach 1:
The patent employs dynamic adaptability through the wavelet transform, where the analysis automatically adjusts to the local characteristics of the signal at each scale. The wavelet coefficients adaptively capture transient features and varying periodicities without requiring preset knowledge of signal frequency or period. This allows the system to handle varying physiological conditions (e.g., changing heart rate, breathing patterns) effectively.
Solution Approach 2:
The patent changes the analysis parameters by using multiple wavelet scales instead of a fixed frequency assumption. Each scale corresponds to different periodicity ranges, and the transform automatically emphasizes the relevant scales based on the actual signal content. This parameter transformation from fixed frequency to multi-scale analysis enables the system to adapt to varying physiological conditions without requiring prior periodicity knowledge.
3Reliability
If noise filtering is applied to improve signal quality, then the measurement reliability improves, but the loss of information may occur in the filtering process
Solution Approach 1:
The patent applies local quality enhancement by processing different scale components with different strategies. Each wavelet scale represents a specific frequency band with different noise characteristics. The method preserves signal information at scales where physiological features are present while filtering noise at other scales, thereby improving reliability without uniform information loss. The scalogram visualization helps identify which scales contain meaningful physiological information versus noise.
Solution Approach 2:
The wavelet transform acts as an intermediary between the raw signal and the final physiological parameter extraction. It provides a intermediate representation (wavelet coefficients and scalogram) that separates signal from noise across different scales. This intermediary representation allows selective processing - preserving important physiological information while removing noise - before final parameter calculation, thus improving reliability with minimal information loss.
Data Source
AI summary
Methods and systems are disclosed for producing a plurality of archetype signals in wavelet space at a plurality of wavelet scales. A signal is transformed using a continuous wavelet transform based at least in part on a wavelet function. A scale dependent archetype transformed signal is computed based at least in part on the transformed signal and based at least in part on a natural periodicity of the wavelet function used to transform the signal. Information may be derived about the signal from the archetype transform signal, and stored in memory.