NIR Spectroscopy Baseline Drift Removal for Blood Compound Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for non-invasive monitoring of blood compound concentrations using Near-Infrared (NIR) spectroscopy face challenges in removing baseline drift and extracting effective features for accurate prediction, especially when dealing with interfering compounds.
Innovation Solution
The method involves removing baseline drift from NIR spectroscopy data using Principal Component Analysis (PCA), obtaining drift-free spectral features, and then extracting a set of global features for regression-based concentration estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If NIR spectroscopy is used for non-invasive monitoring of blood compound concentration, then the monitoring can be performed without blood sampling, but baseline drift adversely affects prediction accuracy
Solution Approach 1:
The patent extracts and removes the baseline drift component from the NIR spectroscopy data using algorithms such as asymmetric least squares (ALS) or polynomial fitting. By separating and eliminating the drift component, the method preserves the non-invasive advantage while improving prediction accuracy of blood compound concentrations.
Solution Approach 2:
The patent applies parameter changes by transforming the raw NIR spectral data through various preprocessing techniques including drift removal algorithms, normalization, and scattering correction. These parameter transformations enhance the quality of spectral features used for concentration prediction.
2Measurement precision
If drift removal algorithms are applied to NIR spectroscopy data, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies partial action by selecting specific drift removal techniques appropriate for different scenarios. Instead of always applying complex algorithms, the method uses simpler approaches when sufficient and reserves more complex algorithms like ALS for cases where higher precision is needed, balancing computational cost with accuracy requirements.
Solution Approach 2:
The patent performs drift removal as a preliminary step before concentration prediction. By preprocessing the NIR data to eliminate baseline drift beforehand, the subsequent prediction algorithms work with cleaner data, reducing their computational burden and improving overall efficiency.
3Productivity
If features are extracted from NIR spectroscopy data for regression analysis, then concentration prediction can be performed, but interfering compounds reduce prediction accuracy
Solution Approach 1:
The patent introduces intermediary processing steps between raw NIR data and regression analysis. These include drift removal, scattering correction, and selective feature extraction that act as mediators to isolate the spectral features related to the target blood compound while filtering out interference from other compounds present in the sample.
Solution Approach 2:
The patent segments the NIR spectral data into distinct feature regions that are specifically relevant to the target blood compound. By focusing on particular wavelength ranges and spectral features associated with the compound of interest, the method reduces the impact of interfering compounds and improves prediction accuracy.
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 effectively removes baseline drift and extracts relevant features, improving the accuracy of blood compound concentration prediction in non-invasive NIR spectroscopy applications.
Implementation Method 1
Monitoring of concentration of blood compounds has always been a topic of much interest. The monitoring of concentration of the blood compounds are typically performed invasively wherein the skin of a test (human or animal) subject is pierced to obtain a blood sample for testing. In a non-invasive method, collection of blood sample is not required for prediction of concentration of the blood compound. Some of the typical methods used for monitoring the concentration of blood compounds non-invasively are Mid-Infrared (Mid-IR), Near-Infrared (NIR), and Raman spectroscopy.
Data Source
Figure 1A
Figure 1B
Figure 2A
AI summary
A method of estimating concentration of a blood compound may include: removing a baseline drift from Near-Infrared (NIR) spectroscopy data to obtain drift-free spectral features; obtaining a set of global features based on the drift-free spectral features; and estimating the concentration of the blood compound by regression using the set of global features.