Eigen Regression Filter for Optical Flow Image Enhancement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optical imaging techniques like OMAG and LSI struggle to differentiate real blood flow from static tissue and noise components, making it difficult to visualize slow blood flows and quantify capillary blood flow accurately.
Innovation Solution
A method that generates a vector from optical image scans comprising tissue, flow, and noise components, using eigen regression filters to isolate and remove static tissue components, thereby enhancing the quality of flow images and quantifying blood perfusion by calculating mean capillary flow velocity and heterogeneity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical imaging techniques (OMAG, LSI) are used to image blood flow, then flow information can be obtained, but static tissue components and noise cannot be differentiated from real blood flow
Solution Approach 1:
The patent segments the optical imaging signals into distinct components (static tissue, flow, noise) using eigen decomposition. By treating the signal as a sum of independent Gaussian processes, the method separates blood flow information from static tissue and noise, allowing accurate differentiation of true flow signals from false flow artifacts.
Solution Approach 2:
The patent extracts flow components from the mixed optical signals by applying eigen regression filters. The method isolates the flow signal by removing static tissue components and noise through mathematical filtering, thereby extracting pure blood flow information from the composite signal containing multiple interfering elements.
2Measurement precision
If conventional OMAG or LSI methods are used, then imaging can be performed, but visualization of slow blood flows is difficult due to background tissue components
Solution Approach 1:
The patent converts the harmful effect of background tissue components into a beneficial filtering process. By modeling the background tissue as a static component in the eigen decomposition, the method uses the presence of tissue signals to identify and remove them, thereby enhancing the visibility of slow blood flows that would otherwise be obscured by tissue interference.
3Measurement precision
If repeated B-scans are acquired to improve flow detection, then more data is obtained, but computational time increases
Solution Approach 1:
The patent changes the computational approach by applying eigen decomposition and eigen regression filtering that efficiently process repeated B-scans. This mathematical transformation reduces computational complexity compared to conventional methods, allowing fast processing of multiple scans while maintaining high flow detection accuracy and enabling real-time implementation.
Data Source
AI summary
Systems and methods for enhancing quality of a flow image of a sample are provided. An image is obtained from a plurality of optical image scans of a sample comprising blood perfused tissue. A vector comprising tissue components, flow components, and noise is then generated, and eigenvectors and eigenvalues are estimated from the vector. From the eigenvectors and eigenvalues, an eigen regression filter is applied to isolate flow components from the tissue components in the sample. The isolated tissue components may then be removed from the image to enhance visualization and quantification of a flow of dynamic moving particles within the sample.