Edge Improvement in De-noised Reconstructed Medical Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Iterative reconstruction and image domain de-noising algorithms often leave noise at the edges of structures in images, leading to jagged appearances that can mislead radiologists and potentially result in misdiagnosis, especially when smooth edges are incorrectly rendered as rough.
Innovation Solution
A method that involves generating an edge map from de-noised reconstructed image data, calculating noise image data by subtracting de-noised data from original data, multiplying this noise with the edge map to create noisy edge data, and then adding a weighted or unweighted product of this data back to the de-noised image to smooth out edge appearances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If iterative reconstruction or image domain de-noising algorithms are used to reduce noise in reconstructed image data, then noise in flat regions is removed or greatly reduced, but noise remains at the edges of structures causing jagged appearances
Solution Approach 1:
The patent applies different noise handling strategies to different regions of the image. Edge regions receive specialized processing that preserves noise to maintain smooth appearances, while flat regions undergo aggressive noise removal. This local differentiation resolves the contradiction by allowing noise reduction where it improves quality without compromising edge smoothness.
Solution Approach 2:
The image is segmented into edge regions and non-edge regions using edge detection algorithms. This segmentation allows the system to apply different processing operations to each segment - noise removal for non-edge regions and noise preservation for edge regions - thereby resolving the contradiction between overall noise reduction and edge smoothness maintenance.
2Reliability
If edge-preserving penalty functions are used in iterative reconstruction, then noise removal is improved, but edges appear rough or jagged
Solution Approach 1:
Instead of directly preserving edges during reconstruction, the patent inverts the approach by first applying standard noise reduction, then detecting edges, and finally adding back noise specifically to edge regions. This inverted strategy resolves the contradiction by correcting the edge appearance problem after noise reduction has already improved overall image quality.
Solution Approach 2:
The patent extracts noise from the de-noised image by subtracting it from the original noisy image. This extracted noise is then selectively re-added to edge regions, separating the noise handling for edges from the rest of the image and resolving the contradiction between noise reduction and edge smoothness.
Data Source
AI summary
A system includes obtaining de-noised reconstructed image data and edge improving a sub-set of the de-noised reconstructed image data corresponding to edges of structure represented in the de-noise reconstructed image data. A system (100) includes an edge detector (202) that detects an edge map of edge locations within de-noised reconstructed image data, a noise image data generator (204) that generates noise image data by subtracting the reconstructed image data by the de-noised reconstructed image data, a noisy edge image data generator (206) that generates noisy edge image data by multiplying the noise image data and the edge map, and an edge improver (208) that generates edge improved de-noised image data by adding the noisy edge image data and a product of a weight and the de-noised reconstructed image data.