Infrared Vessel Segmentation for Accurate Non-Invasive Analyte Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-invasive analyte testing methods, such as fluorescence and Raman spectroscopy, face challenges in accurately distinguishing between areas with and without blood vessels, leading to inaccurate results due to mixed spectral signals and interference from skin components, and require expensive laboratory-level equipment.
Innovation Solution
An image processing method using infrared and ultraviolet light to distinguish between areas with and without blood vessels, followed by fluorescence spectroscopy to obtain accurate spectral data, utilizing a convolutional neural network model for analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If spectral data collection is performed directly on the image including both blood vessel and non-blood vessel areas, then the testing process is simple, but the test result accuracy deteriorates due to mixed spectral signals
Solution Approach 1:
The patent segments the imaging area into multiple regions based on grayscale thresholds, identifying blood vessel areas versus non-blood vessel areas. This segmentation allows separate spectral data collection from each region type, enabling accurate glucose concentration measurement from non-blood vessel areas while maintaining simple automated processing through threshold-based region classification.
2Measurement precision
If Raman spectroscopy is used to measure blood glucose concentration, then measurement accuracy is improved, but device complexity and cost increase due to requiring laboratory-level equipment
Solution Approach 1:
The patent employs fluorescence spectroscopy with inexpensive light sources and detectors instead of complex Raman spectroscopy systems. This substitution maintains measurement capability while dramatically reducing device complexity and cost, enabling portable and accessible glucose monitoring without requiring laboratory-level equipment.
3Productivity
If spectral signals from different components are mixed together, then data collection efficiency is improved, but the ability to extract blood glucose spectral signal deteriorates
Solution Approach 1:
The patent segments the imaging area into blood vessel and non-blood vessel regions using grayscale thresholding. By collecting spectral data separately from each segmented region, the system maintains data collection efficiency through automated region classification while improving spectral signal extraction accuracy by isolating the non-blood vessel area signal where glucose measurement is most accurate.
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, non-invasive analyte testing with reduced interference, achieving miniaturization and real-time results, and eliminating the need for electrochemical reactions.
Implementation Method 1
The image can indicate distribution data and spectral data in the imaging area of a reflection signal or a fluorescence signal generated by the analyte when irradiated by light
Implementation Method 2
The image can indicate distribution data and spectral data in the imaging area of a reflection signal or a fluorescence signal generated by the analyte when irradiated by light
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The present invention provides an image processing method and system in analyte testing, a medium, and a device. The method includes: image obtaining: obtaining an infrared grayscale image in a first area; first target selection: converting the infrared grayscale image into a two-dimensional matrix, recording a pre-selected first target area, and selecting maximum values and minimum values of a horizontal coordinate and a vertical coordinate: xmax, xmin, ymax, ymin; searching area determining: using xmax+m, xmin-m, ymax+m, ymin-m as searching area boundaries, where m is a preset expansion distance, to obtain searching areas; and second target selection: sequencing the searching areas based on grayscales in descending order, and selecting a top preset quantity of coordinates with greatest grayscales as a second target area.