Curved Multiplanar Reformatting for Nerve Segmentation in MRI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for nerve segmentation in MRI images, such as region-growing and active contouring, face challenges in accurately segmenting highly curved nerves due to low contrast with surrounding anatomical features, limiting their reliability and effectiveness.
Innovation Solution
A method using curved multiplanar reformatting (CMPR) to generate orthogonal planes from a selected cross-sectional image, allowing for the estimation and adjustment of nerve contours, and subsequent surface rendering to segment the nerve volume from surrounding structures, enhancing visualization and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If region-growing techniques are used to segment nerves, then the segmentation process can be automated, but the method fails to accurately segment nerves with low contrast against surrounding anatomical features
Solution Approach 1:
The patent divides the segmentation task into multiple stages: initial contour estimation using active contouring, user refinement of contour points, and volumetric segmentation. This multi-stage approach allows automated processing while maintaining accuracy through user feedback at critical stages.
Solution Approach 2:
The patent incorporates user feedback mechanisms where operators can adjust and refine automatically estimated nerve contours. This feedback loop allows correction of automated segmentation errors in low-contrast regions, improving overall segmentation accuracy while maintaining automation benefits.
2Measurement precision
If active contouring techniques are used to segment nerves, then some low contrast issues are overcome, but the method still has lower reliability in low contrast areas and requires proper initialization
Solution Approach 1:
The patent performs preliminary actions by generating multiple planar reformatted images from the 3D volume before final segmentation. These pre-generated images provide enhanced contrast and multiple viewing angles, making the subsequent active contouring more reliable even in challenging low-contrast regions.
Solution Approach 2:
The patent transforms the 2D active contouring problem into a 3D volumetric segmentation problem. By generating planar reformats and then performing volumetric segmentation, the method leverages the third dimension to overcome limitations of 2D contouring in low-contrast areas, improving reliability without sacrificing precision.
3Productivity
If existing nerve segmentation methods are used, then segmentation can be performed, but highly curved nerves such as those in the brachial plexus are difficult to segment accurately
Solution Approach 1:
The patent employs dynamic planar reformatted images that can be generated from multiple orientations. This dynamic approach allows the segmentation algorithm to adapt to curved nerve paths by examining the nerve from different planar angles, improving accuracy for highly curved nerves while maintaining efficient processing speeds.
Solution Approach 2:
The patent transitions from 2D planar images to 3D volumetric segmentation. This phase transition enables accurate representation and segmentation of highly curved nerves by capturing their three-dimensional geometry, overcoming the limitations of 2D-based methods while maintaining computational efficiency.
4Measurement precision
If curved multiplanar reformatting is used to generate orthogonal planes, then three-dimensional nerve visualization is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the complex computational task into distinct modules: generating planar reformats, estimating nerve contours in each plane, refining contours with user feedback, and performing volumetric segmentation. This segmentation of the processing pipeline reduces overall computational complexity while maintaining high visualization accuracy.
Solution Approach 2:
The patent generates multiple planar reformatted images from the 3D volume, creating more data than strictly necessary. This excessive action provides redundant information that improves the reliability of nerve contour estimation and visualization accuracy, while the computational cost is managed through efficient algorithms and selective processing.
Data Source
AI summary
Systems and methods for segmenting a nerve in a three-dimensional image volume obtained with a magnetic resonance imaging (“MRI”) system are provided. A three-dimensional image volume that depicts a nerve and surrounding anatomical structures is provided and from that image volume the nerve is segmented. In general, a curved multiplanar reformatting (“CMPR”) process is utilized to mark, segment, and then display the nerve in three dimensions.