Physiological Model-Based Medical Image Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image registration techniques face challenges in multi-modal registration of higher dimensional images due to complexity in mutual intensity information, leading to long processing times and poor results, limiting their adoption in clinical workflows.
Innovation Solution
The method involves fitting a physiological model of an anatomical structure learned from a database to multiple images using discriminative machine-learning techniques, such as marginal space learning, to generate correspondences and register images, thereby exploiting high-level prior knowledge for accurate alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional image-based registration techniques are used for multi-modal registration of higher dimensional images, then the registration can be performed using general-purpose methods, but the processing time becomes very long and the results are poor due to the complexity of mutual intensity information
Solution Approach 1:
The patent introduces a physiological model as an intermediary representation that mediates between multiple medical images from different modalities. Instead of directly registering complex multi-modal images using mutual intensity information, the method fits a physiological model to each image independently, then uses the fitted model parameters for registration. This intermediary model simplifies the registration problem by transforming high-dimensional image data into lower-dimensional physiological parameters, thereby reducing processing time while maintaining adaptability to multi-modal images
Solution Approach 2:
The patent extracts essential physiological information from complex multi-modal medical images by fitting a physiological model to each image. This extraction process isolates the key anatomical and physiological characteristics from the complex image data, creating a simplified representation that can be efficiently registered. By taking out only the essential physiological features rather than processing the entire complex image data, the method reduces computational burden and processing time
2Adaptability or versatility
If conventional image-based registration techniques are used for multi-modal registration of higher dimensional images, then the registration can be performed using general-purpose methods, but the alignment accuracy becomes poor due to the complexity of mutual intensity information
Solution Approach 1:
The physiological model serves as an intermediary that improves alignment accuracy by providing a unified anatomical framework across different imaging modalities. The model fitting process captures essential physiological structures and their relationships, enabling more accurate correspondence establishment between images. This intermediary representation transforms the difficult problem of directly comparing complex multi-modal images into a simpler problem of aligning physiological parameters, thereby improving measurement precision
Solution Approach 2:
The patent transforms the registration problem from operating on raw image intensity values to operating on physiological model parameters. By changing the parameter space from complex multi-dimensional image data to reduced-dimensional physiological characteristics, the method improves alignment accuracy. The physiological parameters provide a more meaningful and stable basis for registration compared to direct intensity-based comparisons, leading to better correspondence between anatomical structures across different modalities
Data Source
AI summary
A method and system for physiological image registration and fusion is disclosed. A physiological model of a target anatomical structure in estimated each of a first image and a second image. The physiological model is estimated using database-guided discriminative machine learning-based estimation. A fused image is then generated by registering the first and second images based on correspondences between the physiological model estimated in each of the first and second images.