Dual-Energy CT Image Segmentation via Smoothness Prior
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current segmentation algorithms in non-invasive diagnostic imaging, such as CT scans, face challenges in accurately identifying organ boundaries due to noise and intensity variations, which hinders automated segmentation and affects image quality.
Innovation Solution
A method involving dual-energy imaging, where a first image is reconstructed for diagnostics using a normal iterative reconstruction algorithm and a second image is reconstructed with a modified algorithm prioritizing smoothness over textural details, allowing for accurate segmentation of the first image based on the segmentation of the second image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If iterative reconstruction (IR) algorithms are used to reduce noise levels, then noise reduction is improved, but image intensity uniformity deteriorates due to variations in intensity within anatomical structures
Solution Approach 1:
The patent segments the image processing task into two distinct stages: first reconstructing an image with noise reduction properties, then performing intensity normalization specifically within segmented anatomical structures. This segmentation allows different processing strategies to be applied to different aspects of image quality without interference.
Solution Approach 2:
The patent performs preliminary segmentation of anatomical structures before final image output. By identifying and isolating anatomical regions in advance, the system can apply targeted intensity normalization to these regions, ensuring uniform intensity within each structure while preserving the noise reduction benefits of iterative reconstruction.
2Productivity
If automatic segmentation is performed on images with intensity variations, then segmentation speed is improved, but segmentation accuracy deteriorates due to noise and intensity fluctuations
Solution Approach 1:
The patent performs preliminary intensity normalization within anatomical structures before segmentation is finalized. By pre-processing the intensity distribution in key anatomical regions, the system creates more favorable conditions for automatic segmentation algorithms, improving their accuracy without requiring manual intervention.
Solution Approach 2:
The patent implements a feedback mechanism where segmentation results are used to identify anatomical structures, which then guide intensity normalization, which in turn improves subsequent segmentation. This iterative feedback loop continuously refines both the segmentation accuracy and the intensity uniformity based on the most current information.
Data Source
AI summary
Methods and systems are provided for reconstructing and automatically segmenting an image. In one embodiment, a method comprises acquiring projection data, the projection data comprising higher energy projection data and lower energy projection data, generating a first image from the projection data, generating a second image from the projection data, segmenting the second image to generate segments, and segmenting the first image based on the segments of the second image. In this way, an image which may otherwise prove challenging for an automatic segmentation process may be accurately segmented without sacrificing textural details of the image.