U-Net Neural Network for OCTA Flow Artifact Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for correcting flow artifacts in optical coherence tomography angiography (OCTA) are either slow or dependent on segmentation, limiting their effectiveness and efficiency.
Innovation Solution
A neural network approach is used to correct flow artifacts in OCTA volumes, which is faster than traditional methods, independent of slab definitions, and capable of easy parallelization, allowing for efficient processing of large 3D data arrays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If slab-based methods are used to correct projection artifacts, then artifact correction can be achieved, but processing speed is slow and segmentation dependency increases complexity
Solution Approach 1:
The patent replaces traditional slab-based mechanical processing methods with a neural network approach. The neural network directly processes 3D OCTA volumes to correct flow projection artifacts, eliminating the need for manual segmentation and iterative slab processing, thereby significantly improving processing speed while maintaining correction accuracy.
Solution Approach 2:
The patent changes the fundamental parameter of processing from 2D slab-based iteration to 3D volume-based single-pass processing. By training the neural network to directly output corrected flow images from input OCTA volumes, the system achieves faster processing without sacrificing artifact correction reliability.
2Reliability
If slab-based methods are used for artifact correction, then flow artifacts can be reduced, but dependency on segmentation definitions increases device complexity
Solution Approach 1:
The patent replaces segmentation-dependent mechanical processing with a neural network system that directly processes 3D OCTA volumes. The neural network learns to identify and correct flow projection artifacts without requiring manual or automated segmentation definitions, thereby reducing device complexity while maintaining correction reliability.
Solution Approach 2:
The neural network system provides universal artifact correction across different retinal layers and vessel configurations without requiring separate segmentation definitions for each case. The single trained model handles various scenarios, eliminating the need for complex, case-specific segmentation pipelines.
3Adaptability or versatility
If volume-based methods are used to correct artifacts, then processing can be done in 3D space, but processing time increases due to large data array analysis
Solution Approach 1:
The patent replaces computationally intensive volume-based algorithms with a neural network approach. The neural network processes 3D OCTA volumes efficiently by learning direct mappings from input to corrected output, achieving fast processing times while maintaining full 3D processing capability and visualization versatility.
Data Source
AI summary
A system and/or method uses a trained U-Net neural network to remove flow artifacts from optical coherence tomography (OCT) angiography (OCTA) data. The trained U-Net receives as input both OCT structural volume data and OCTA volume data, but expands the OCTA volume data to include depth information. The U-Net applies dynamic pooling along the depth direction and weighs more heavily the portion of the data that follows (e.g., along the contours of) select retinal layers. In this manner the U-Net applies contextually different computations at different axial locations based at least in part on the depth index information and/or (e.g., proximity to) the select retinal layers. The U-net outputs OCT volume data of reduced flow artifacts as compared with the input OCTA data.