Liver Segmentation in CT via Hierarchical Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for liver segmentation from abdominal CT scans are tedious and time-consuming due to challenges posed by intensity inhomogeneity and similarities within and outside the liver, requiring more accurate and automated approaches beyond simple intensity differences.
Innovation Solution
A method involving a flowchart of four steps: rough region estimation, rough segmentation, coarse segmentation, and fine segmentation, utilizing anatomical knowledge, intensity thresholds, graph-cut seeds, fast Gaussian Mixture Model, and shape models to achieve accurate and efficient liver segmentation with minimal human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual delineation by radiologists is used, then segmentation accuracy can be maintained, but the process becomes tedious and time-consuming
Solution Approach 1:
The patent divides liver segmentation into multiple stages: rough region estimation, rough segmentation, coarse segmentation, and fine segmentation. Each stage progressively refines the segmentation result, automating the process while maintaining accuracy by breaking down the complex task into manageable steps that can be performed algorithmically
Solution Approach 2:
The patent performs preliminary rough region estimation and rough segmentation using intensity thresholds and anatomical knowledge before proceeding to more detailed segmentation. This preliminary action prepares the data structure and identifies key regions, enabling subsequent automated processing to achieve final segmentation accuracy without requiring manual intervention at each step
2Productivity
If simple intensity differences are used for segmentation, then the process becomes simpler and faster, but accuracy deteriorates due to intensity inhomogeneity and similarities
Solution Approach 1:
The patent applies different segmentation strategies to different regions of the liver based on local characteristics. The rough segmentation phase uses intensity thresholds for speed, while the fine segmentation phase uses more sophisticated methods like graph-cut and shape models specifically at boundary regions where accuracy is critical, thus achieving both speed and precision
Solution Approach 2:
The patent transforms the segmentation problem from simple intensity-based classification to a multi-parameter optimization problem. By introducing shape constraints, anatomical priors, and hierarchical refinement parameters, the system overcomes the limitations of simple intensity differences while maintaining computational efficiency through progressive refinement
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for automatic organ segmentation in CT images comprises in a first step, rough region segmentation; in a second step, coarse organ segmentation; and in a third step, refinement of organ segmentation. The organ may be a liver. Rough region segmentation may comprise applying standard anatomical knowledge to the CT images. Coarse segmentation may comprise identifying organ voxels using a probabilistic model. Refinement of organ segmentation may comprise refinement based on intensity, followed by refinement based on shape. Apparatuses configured to carry out the method are also disclosed.