Iterative Image Registration with Volume-Based Deformation Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image registration and segmentation algorithms are prone to underperformance due to imperfections in input imagery, leading to sub-optimal accuracy, especially in clinical applications where precise alignment is crucial for subsequent processing steps.
Innovation Solution
An image processing system that includes an iterative solver driven by an optimization function with a similarity measure, a volume monitor to assess neighborhood volume changes, and a corrector to mitigate the effect of the similarity measure when volume changes violate an acceptance condition, ensuring robustness against artifacts and anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a standard iterative registration algorithm driven by similarity measure is used, then the registration process is simple and fast, but the accuracy deteriorates when input images contain imperfections or artifacts
Solution Approach 1:
The patent applies preliminary action by performing a first registration pass to obtain an initial transformation, then using this result to guide a second refined registration pass. This preliminary registration step prepares the data for more accurate subsequent processing, improving final accuracy without completely redesigning the entire algorithm.
Solution Approach 2:
The registration process is segmented into multiple distinct passes: a first registration pass using the objective function, followed by a second registration pass that refines the transformation. This segmentation allows each pass to have specialized functionality, with the second pass focusing on correcting artifacts while preserving anatomical features from the first pass.
2Measurement precision
If the similarity measure is applied strongly to achieve precise alignment, then registration accuracy improves, but unwanted deformations and artifacts are amplified
Solution Approach 1:
The patent applies partial action by using the similarity measure-driven registration in a controlled, partial manner during the second pass. Rather than fully trusting the similarity measure, the algorithm selectively applies its guidance while constraining deformations through the transformation model, achieving precise alignment without excessive deformation amplification.
Solution Approach 2:
The system uses feedback by evaluating the transformation from the first pass and using this information to guide the second registration pass. The feedback mechanism allows the algorithm to adjust the application of the similarity measure, preventing harmful deformations while maintaining alignment precision through iterative refinement.
3Adaptability or versatility
If non-rigid registration is performed to handle anatomical variations, then adaptability improves, but computational time and complexity increase
Solution Approach 1:
The patent applies periodic action by structuring the registration as discrete, periodic passes rather than continuous optimization. The first pass performs coarse alignment, then the second pass performs refined alignment. This periodic structure handles anatomical variations adaptively while controlling computational time through clear termination criteria for each pass.
Solution Approach 2:
The first registration pass serves as preliminary action that handles the bulk of anatomical variation compensation. By pre-processing the alignment in this initial pass, the second pass only needs to perform fine-tuning, significantly reducing the computational burden while maintaining high adaptability to anatomical variations.
Data Source
AI summary
Systems and methods for iteratively computing an image registration or an image segmentation are driven by an optimization function that includes a similarity measure component whose effect on the iterative computations is relatively mitigated based on a monitoring of volume changes of volume elements at image locations during the iterations. A system and a related method quantify a registration error by applying a series of edge detectors to input images and combining related filter responses into a combined response. The series of filters are parameterized with a filter parameter. An extremal value of the combined response is then found and a filter parameter associated with said extremal value is then returned as output. This filter parameter relates to a registration error at a given image location.