Remote PPG Signal Extraction Using Coefficient Selection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Remote photoplethysmography (PPG) measurements face challenges in signal detection and processing due to noise, distortion, and specular reflections, especially in unobtrusive setups where signal-to-noise ratio is low and subject movement occurs, making it difficult to accurately extract vital sign data.

Innovation Solution

A device and method that utilize a transforming unit to determine coefficient components from wavelength components of a data stream, allowing for the selection of a single coefficient component using a predetermined reference vector, which minimizes computational resources and time, and ensures deterministic signal component selection, thereby improving the efficiency and accuracy of extracting physiological information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If remote photoplethysmography measurements are performed in unobtrusive setups, then subject freedom of movement and comfort are improved, but signal-to-noise ratio deteriorates and measurement precision worsens

Engineering Contradiction:
Improvesubject freedom of movementVSAvoidsignal-to-noise ratio
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent segments the detected electromagnetic radiation signal into multiple wavelength components, then further decomposes these into independent source signals using Independent Component Analysis. This segmentation allows the system to isolate the physiological signal from noise and interference, maintaining measurement precision despite the unobtrusive remote setup.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary processing stage that uses a predetermined reference vector to guide the selection of coefficient components during Independent Component_analysis. This intermediary mechanism deterministically identifies the correct physiological signal component among multiple decomposed signals, improving measurement precision without requiring obtrusive contact.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If multiple coefficient components are determined and evaluated to select the correct signal component, then signal accuracy is improved, but computational complexity and detection time increase

Engineering Contradiction:
Improvesignal component selection accuracyVSAvoidcomputational resources
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary determination of multiple coefficient components from wavelength components before the actual signal component selection. By pre-calculating these coefficient components and having them ready for evaluation against the reference vector, the system streamlines the selection process and reduces overall computational complexity during detection.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent transforms the original wavelength component data into coefficient components through a mathematical transformation process. This parameter change from wavelength domain to coefficient domain simplifies the subsequent selection process by providing a structured framework for comparing signal components against the reference vector, reducing computational burden.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If deterministic coefficient component selection using a predetermined reference vector is implemented, then signal component selection reliability is improved, but device complexity increases

Engineering Contradiction:
Improvesignal component selection consistencyVSAvoidprocessing algorithm complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The predetermined reference vector serves as a self-contained guide that enables the system to automatically and deterministically identify the correct signal component without requiring complex external validation or manual intervention. The reference vector encapsulates the necessary criteria for signal identification, allowing the processing algorithm to autonomously make reliable selections.

Inventive Principle:
Principle #25Self-service

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

This approach reduces computational demands and detection time, allowing for more efficient and accurate extraction of vital sign data, even in conditions with low signal-to-noise ratios and subject movement, by focusing on the coefficient component with the highest variance or closest to a predetermined reference vector, thus enhancing the reliability of remote PPG measurements.

Implementation Method 1

extracting physiological information from remotely detected electromagnetic radiation emitted or reflected by a subject

Methodology Applied
Scientific EffectElectromagnetic radiation detection:

Data Source

PatentEP4000505A1Device and method for extracting physiological information
Publication Date: 2022.05.25 KONINKLIJKE PHILIPS NV
  • EP4000505A1 patent drawingFigure 1
  • EP4000505A1 patent drawingFigure 2a~2b
  • EP4000505A1 patent drawingFigure 3

AI summary

The present invention relates to a device and a method for extracting physiological information from remotely detected electromagnetic radiation (14) emitted or reflected by a subject. A data stream (24) derived from detected electromagnetic radiation (14) is received, the data stream (24) comprising a sequence of signal samples representing a region of interest (58) exhibiting a continuous or discrete characteristic signal including physiological information indicative of at least one at least partially periodic vital signal (20), and being comprised of at least two wavelength components. The device further comprises a transforming unit for providing coefficient components from the at least two wavelength components and a coefficient selection unit for selecting the coefficient component which is used for calculating the desired signal component used for determining the physiological information.