Anatomical Atlas Generation via Inverse Average Transformations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining anatomical atlas data struggle to accurately represent average anatomical structures and pathological changes across diverse patient images, especially when these images are captured using different parameter sets, leading to inconsistencies and inefficiencies in medical imaging and diagnostics.
Innovation Solution
A data processing method that acquires patient data and model data, applies matching transformations to align patient images with anatomical atlases, and uses inverse average transformations to determine atlas data, which represents an average anatomical structure or pathological changes by averaging and inverting transformations from patient images to model images, thereby improving the accuracy and consistency of anatomical representations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If patient images from different parameter sets are directly used to determine anatomical atlas data, then the method can handle diverse imaging data, but the anatomical representations become inconsistent and imprecise
Solution Approach 1:
The patent applies parameter changes by transforming patient images from different parameter sets into a common reference frame. Matching transformations are computed to adjust images based on their parameter sets, allowing diverse imaging data to be consistently integrated into the anatomical atlas while maintaining precision across different imaging modalities and parameters
Solution Approach 2:
The patent introduces matching transformations as an intermediary mechanism between patient images and the anatomical atlas. These transformations act as mediators that align patient-specific images with the general anatomical structure, enabling consistent integration of diverse imaging data without compromising anatomical representation precision
2Productivity
If model data is used as a starting point for determining atlas data, then the process can be efficient, but the initial model may not accurately represent average anatomical structures across diverse patients
Solution Approach 1:
The patent applies preliminary action by using model data as an initial starting point that represents a general anatomical structure. This preliminary model is then systematically improved by incorporating patient-specific data through matching transformations, allowing efficient progression from a generic model to a precise average anatomical representation across diverse patients
Solution Approach 2:
The patent implements feedback by using patient images to refine and improve the model data. The matching transformations computed from patient images provide feedback that adjusts the initial model, progressively improving the accuracy of the anatomical atlas representation while maintaining computational efficiency
3Measurement precision
If matching transformations are applied to align patient images with anatomical atlases, then anatomical representations become more consistent, but the computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the complex matching transformation into manageable components. The transformation is segmented into parameter-set-specific adjustments and common anatomical alignment operations, reducing computational complexity while maintaining consistency across diverse patient images and the anatomical atlas
Data Source
AI summary
A data processing method for determining data which are referred to as atlas data and comprise information on a description of an image of a general anatomical structure, wherein this image is referred to as the atlas image, the method comprising the following steps performed by a computer: acquiring patient data which comprise a description of a set of images of an anatomical structure of a set of patients, wherein the images are referred to as patient images and each patient image is associated with a parameter set which comprises one or more parameters which obtain when the patient images are generated, wherein the parameters influence representations of anatomical elements as expressed by image values in the patient images; acquiring model data which comprise information on a description of an image of a model of an anatomical structure of a (single or average or generic) patient which is referred to as the model image and is associated with the parameter set; determining matching transformations which are referred to as PM transformations and which are constituted to respectively match the set of patient images of the set of patients to the model image by matching images associated with the same parameter set; determining an inverse average transformation by applying an inverting and averaging operation to the determined PM transformations; and a) determining the atlas data by applying the determined inverse average transformation to the model data; or b) respectively applying the determined PM transformations to the respective patient images in order to determine matched patient images, averaging the matched patient images in order to determine an average matched patient image, and determining the atlas data by applying the determined inverse average transformation to the average matched patient image.