Speckle Reduction in Optical Coherence Tomography via Matrix Completion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current OCT imaging systems, such as the Topcon DRI OCT-1 swept source OCT, suffer from significant speckle noise in single frames, which limits image quality due to the need for averaging over a small number of neighboring frames, leading to potential blurring when more frames are included, especially in rapidly changing retinal structures.
Innovation Solution
The method involves reconstructing each A-scan from a set of neighboring A-scans within a defined two-dimensional region, forming a matrix and performing matrix completion to minimize speckle noise, using either a low-rank matrix or a combination of low-rank and sparse matrices to reduce noise while maintaining image sharpness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If moving average is computed over more neighboring frames to reduce speckle noise, then speckle reduction is improved, but image sharpness deteriorates due to blur in rapidly changing regions
Solution Approach 1:
The patent applies local quality by differentiating treatment based on region characteristics. It computes motion vectors to identify regions with high motion (rapidly changing structures) versus low motion regions. Different weighting factors are applied: higher weights for static regions to maximize speckle reduction, lower weights for dynamic regions to preserve sharpness. This localized adaptive approach resolves the contradiction by allowing optimal speckle reduction in appropriate regions while preventing blur in rapidly changing areas.
Solution Approach 2:
The patent implements dynamics through motion vector computation and adaptive weighting. Instead of using a fixed averaging window, the system dynamically adjusts the effective number of frames used in averaging based on local motion characteristics. Regions with high motion vectors receive different processing weights than static regions, allowing the system to adapt the speckle reduction intensity dynamically to local image content and motion state.
2Manufacturing precision
If moving average is computed over a small number of neighboring frames, then image sharpness is preserved, but speckle reduction is limited
Solution Approach 1:
The patent resolves this contradiction through local quality by applying different averaging intensities to different regions. In static or low-motion regions, the system uses higher averaging weights to maximize speckle reduction. In dynamically changing regions, it uses lower weights to preserve sharpness. This localized differentiation allows the system to achieve strong speckle reduction where appropriate while maintaining sharpness where needed, overcoming the limitation of small-frame averaging.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the effective number of frames (k) used in moving average computation based on local motion characteristics. Instead of using a fixed small k value, the system computes motion vectors and adapts k locally - using larger effective k values in static regions for better speckle reduction, and smaller values in dynamic regions to preserve sharpness. This parameter adaptation resolves the contradiction between sharpness and speckle reduction.
3Object-affected harmful factors
If conventional filters are applied to reduce speckle, then speckle noise is reduced, but edge sharpness deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the image into different regions based on motion characteristics. It computes motion vectors for each pixel or region and segments the image into high-motion and low-motion regions. Different filtering strategies are then applied to each segment: aggressive speckle reduction in static regions, and preserved sharpness in dynamic regions. This segmentation approach allows conventional filtering to work effectively in appropriate regions while protecting edges in dynamic regions.
Solution Approach 2:
The patent implements dynamics through motion-adaptive filtering. Instead of applying a fixed filter strength throughout the image, the system dynamically adjusts filtering intensity based on local motion vectors. Regions with high motion (likely containing edges or rapidly changing structures) receive reduced filtering intensity to preserve sharpness, while static regions receive strong filtering for optimal speckle reduction. This dynamic adaptation resolves the contradiction between speckle reduction and edge preservation.
Data Source
AI summary
An optical coherence tomography (OCT) image composed of a plurality of A-scans of a structure is analyzed by defining, for each A-scan, a set of neighboring A-scans surrounding the A-slices scan. Following an optional de-noising step, the neighboring A-scans are aligned in the imaging direction, then a matrix X is formed from the aligned A-scans, and matrix completion is performed to obtain a reduced speckle noise image.