Head Optical Signal Estimation for Accurate Cerebral Biometrics

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing biometric monitoring methods, such as MRI and NIRS, face challenges in providing accurate and continuous monitoring of brain-related biometric information due to high costs and low measurement accuracy, respectively, with NIRS being affected by non-cerebral anatomical structures.

Innovation Solution

A machine learning-based method that utilizes an estimation model trained on anatomical and biometric data to analyze optical signals from a person's head, considering the head's structure, to enhance monitoring accuracy and cost-effectiveness.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If NIRS is used for monitoring, then cost-effectiveness and continuous monitoring capability are improved, but measurement accuracy deteriorates due to interference from non-cerebral anatomical structures

Engineering Contradiction:
Improvecontinuous monitoring capabilityVSAvoidbiometric information accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent segments the head anatomical structure into multiple layers (skin, skull, cerebrospinal fluid, gray matter, white matter) and assigns different optical properties to each layer. This segmentation allows the system to distinguish between cerebral and non-cerebral contributions to the NIRS signal, thereby improving measurement accuracy while maintaining continuous monitoring capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by assigning different optical properties (absorption coefficient, scattering coefficient) to different anatomical layers based on their specific characteristics. This enables the system to account for the unique optical properties of each tissue type and accurately isolate cerebral biometric information from the composite NIRS signal

Inventive Principle:
Principle #3Local quality

2Measurement precision

If MRI or CT is used for monitoring, then measurement accuracy is improved, but cost and ability to provide continuous monitoring worsen

Engineering Contradiction:
Improvebiometric information accuracyVSAvoidcontinuous monitoring capability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent uses NIRS as an intermediary technique that bridges the gap between the high accuracy of MRI/CT and the need for continuous monitoring. By incorporating anatomical structure information into the NIRS analysis, the system achieves cerebral-specific biometric monitoring at a lower cost and with continuous capability

Inventive Principle:
Principle #24Intermediary (Mediator)

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

Enables accurate and continuous monitoring of biometric information, improving the cost-effectiveness of brain-related health assessments by mitigating the impact of non-cerebral anatomical structures on measurement accuracy.

Implementation Method 1

Near-infrared spectroscopy (NIRS) that has been introduced recently is a method of indirectly analyzing a bioactivity occurring in a body portion (e.g., brain or the like) of the person by measuring a degree of attenuation of near-infrared ray (due to scattering and absorption by oxidized or non-oxidized hemoglobin)

Methodology Applied
Scientific EffectNear-infrared spectroscopy: Absorption (EM radiation)

Implementation Method 2

estimating biometric information about the head of the person to be measured by analyzing the acquired analysis target optical signal using an estimation model learned based on data about an anatomical structure of at least one head portion

Methodology Applied
Scientific EffectMachine learning:

Data Source

PatentEP3954276B1Method, system, and non-transitory computer-readable recording medium for estimating biometric information about a head using machine learning
Publication Date: 2025.12.03 KOREA ADVANCED INST OF SCI & TECH
  • EP3954276B1 patent drawingFigure 1~2
  • EP3954276B1 patent drawingFigure 3~4(b)
  • EP3954276B1 patent drawingFigure 5(a)~5(b)

AI summary

According to one aspect of the present disclosure, there is provided a method of estimating biometric information about a head using a machine learning, the method including: acquiring an analysis target optical signal detected from a head portion of a person to be measured by at least one optical sensor disposed on the head portion of the person to be measured; and estimating the biometric information about the head of the person to be measured by analyzing the acquired analysis target optical signal using an estimation model learned based on data about an anatomical structure of at least one head portion, data about a biometrical state of the at least one head portion, and data about a optical signal associated with the data about the anatomical structure of at least one head portion and the data about the biometrical state of the at least one head portion.