PPG-to-ECG Reconstruction Using Joint Dictionary Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ECG monitoring devices are restrictive and require user intervention, while PPG offers user-friendly long-term monitoring but lacks accuracy in reconstructing ECG parameters, especially for diverse patient morphologies, and existing dictionary learning methods fail to effectively bridge the low waveform correlation between PPG and ECG signals.
Innovation Solution
A cross-domain joint dictionary learning (XDJDL) method is employed to learn a dictionary pair for ECG and PPG signals, incorporating a stable linear mapping to enhance representation and generalization, and a neural network architecture for ECG inference and CVD diagnosis, optimized for mobile devices with lightweight variants.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ECG monitoring devices are used, then measurement precision is improved, but ease of operation deteriorates due to restrictive requirements and user intervention needs
Solution Approach 1:
The patent creates a computational copy of ECG signals from PPG measurements using dictionary learning and neural networks. Instead of directly measuring ECG with complex equipment, the system captures PPG signals and reconstructs ECG waveforms through learned mappings, achieving ECG-equivalent information with simpler sensors.
Solution Approach 2:
The patent replaces the mechanical/electrical measurement system of ECG electrodes with an optical measurement system using PPG sensors. The physical contact-based ECG measurement is substituted with optical detection of blood volume changes, followed by computational reconstruction of ECG signals.
2Ease of operation
If PPG is used for long-term monitoring, then ease of operation is improved, but measurement precision deteriorates due to low waveform correlation between PPG and ECG
Solution Approach 1:
The patent introduces dictionary learning as an intermediary computational layer between PPG and ECG signals. The dictionary learned from training data acts as a mediator that captures the complex nonlinear relationships between PPG and ECG waveforms, enabling accurate reconstruction despite their low direct correlation.
Solution Approach 2:
The patent transforms the PPG signal representation into a different parameter space using dictionary coefficients and neural network features. By changing the representation parameters from raw PPG waveforms to learned dictionary coefficients, the system can accurately reconstruct ECG signals that have low correlation with the original PPG waveform.
3Ease of manufacture
If existing dictionary learning methods are used, then ease of manufacture is improved, but measurement precision deteriorates due to inability to bridge low waveform correlation between PPG and ECG
Solution Approach 1:
The patent combines multiple computational components into a composite system: dictionary learning for feature extraction, neural networks for nonlinear mapping, and signal reconstruction algorithms. This composite approach leverages the strengths of each component to overcome the limitations of individual methods in bridging the PPG-ECG correlation gap.
Solution Approach 2:
The patent moves the problem from the time-domain waveform correlation to a higher-dimensional feature space using dictionary coefficients and neural network representations. In this elevated dimensional space, the relationship between PPG and ECG becomes more apparent and can be accurately modeled despite low correlation in the original waveform domain.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Achieves accurate ECG reconstruction and CVD diagnosis with high F1 score of 0.96, providing economical, user-friendly, and long-term cardiac monitoring with reduced memory consumption.
Implementation Method 1
photoplethysmogram (PPG) measures the optical response of the blood volume changes at peripheral ends
Data Source
AI summary
Systems, methods, apparatuses, and computer program products for reconstructing electrocardiogram (ECG) waveforms from photoplethysmogram (PPG) signals using dictionary learning and deep learning for continuous monitoring and analytics. A method may include obtaining an electrical signal of a heart. The method may also include obtaining a circulatory signal related to a pulsatile volume of blood in tissue. The method may further include preprocessing the electrical signal and the circulatory signal. In addition, the method may include learning a mapping from at least one of the preprocessed electrical signal and the preprocessed circulatory signal. Further, the method may include training a model using the mapping. The method may also include deriving cardiovascular data based on the trained model.