Feature Space MR-Guided PET Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current PET image reconstruction methods using anatomical priors are vulnerable to mismatches between anatomical images and true activity distribution, leading to potential masking of molecular information or creation of non-existing abnormalities, as they enforce anatomical boundaries in image space.
Innovation Solution
The method employs a penalized maximum-likelihood reconstruction technique that utilizes anatomical information in the feature space instead of image space, combining relative difference priors from both spaces to improve spatial resolution and signal-to-noise ratio, specifically using MR-guided block sequential regularized expectation maximization (MRgBSREM) with PET/MR scanners.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If anatomical information is used in image space to enforce similarity between neighboring voxels, then spatial resolution is improved, but mismatches between anatomical images and true activity distribution occur leading to masking of molecular information or creation of non-existing abnormalities
Solution Approach 1:
The patent transforms the penalty calculation from image space to feature space by mapping voxel intensities across multiple anatomical images into a multidimensional feature vector. This dimensional transformation allows the penalty function to operate on extracted features rather than raw image intensities, enabling better discrimination between true anatomical boundaries and mismatches while preserving molecular information integrity
2Manufacturing precision
If anatomical boundaries are enforced in image space, then image quality is improved, but PET information may be masked or non-existing abnormalities may be created
Solution Approach 1:
The patent implements local quality by calculating penalties based on feature-space distances for each voxel independently, allowing different regions to have different penalty strengths based on their local feature characteristics. This enables the method to preserve molecular information in regions where anatomical boundaries do not align with true activity distributions while still enforcing anatomical constraints where appropriate
Data Source
AI summary
A method for PET image reconstruction acquires PET data by a PET scanner; reconstructs from the acquired PET data a seed PET image; builds a feature space from the seed PET image and anatomical images co-registered with the seed PET image; performs a penalized maximum-likelihood reconstruction of a PET image from the seed PET image and the feature space using a penalty function that is calculated based on the differences between each voxel and its neighbors both on the PET image and in the feature space regardless of their location in the image.