Deformity-Weighted Medical Image Registration for Soft Tissue
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for registering pre-operative planning images with intra-operative surface scans are prone to errors due to anatomical differences and non-homogeneous deformations, particularly in softer tissue regions, leading to inaccurate registrations.
Innovation Solution
A method that considers the elasticity and deformability of tissue by using atlas data to assign confidence values based on tissue type, which are then used to weight the registration process, compensating for deformations and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If classical surface registration methods are used without considering tissue elasticity, then the registration process is simple and fast, but registration accuracy deteriorates due to unaccounted deformations in softer tissue regions
Solution Approach 1:
The patent applies different confidence values to different regions of the surface based on tissue elasticity characteristics. Softer tissue regions receive lower confidence values while harder regions receive higher confidence values, allowing the registration to adapt to local variations in tissue properties without requiring a complete redesign of the registration algorithm.
Solution Approach 2:
The patent introduces atlas data as an intermediary that contains pre-computed confidence values for different tissue regions. This intermediary layer provides the necessary tissue elasticity information without requiring direct measurement or complex real-time computation during the registration process itself.
2Reliability
If confidence values based on tissue elasticity are incorporated, then registration accuracy improves by compensating for deformations, but the computational time and processing complexity increase
Solution Approach 1:
The confidence values are pre-computed and stored in the atlas data before the actual registration process. This preliminary action allows the registration system to quickly retrieve and apply appropriate confidence values without performing complex calculations in real-time during surgery.
Solution Approach 2:
The patent uses a statistical atlas that represents averaged tissue elasticity characteristics from multiple subjects. This copying approach allows the system to apply general tissue elasticity patterns to individual patients without requiring patient-specific elasticity measurements, significantly reducing computational requirements.
3Measurement precision
If uniform confidence values are assigned to all surface regions, then the registration process is simple and efficient, but registration accuracy deteriorates due to non-homogeneous tissue deformation
Solution Approach 1:
The patent assigns different confidence values to different surface regions based on their underlying tissue elasticity characteristics. Regions with softer tissue receive lower confidence values while regions with harder tissue receive higher confidence values, allowing the system to account for non-homogeneous deformation patterns.
Solution Approach 2:
The surface is effectively segmented into different tissue types based on the atlas data, with each segment receiving appropriate confidence values. This segmentation allows the registration to treat different regions differently according to their specific deformation characteristics.
Data Source
Figure 1
Figure 2
AI summary
Disclosed is a computer-implemented method of determining a spatial relationship between planning image data and current surface data which leads to improved surface registration accuracy by considering the elasticity and deformability of the tissue. The knowledge about the tissue can be estimated based on type of tissue and atlas information. For the process of generating surface registration points on specific anatomical regions, e.g. the face or forehead, are acquired with a classical navigated pointer or laser pointer. It is also possible to acquire points with surface scanners. Confidence values defining a probability for certain parts of the surface registration points being deformed in comparison to a planning image are read from atlas data and used to compensate for the deformation in the registration between the surface registration points and the planning image in order to render the registration valid.