Edge Preserving Penalized Reconstruction for PET Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In positron emission tomography (PET) imaging, noise suppression techniques often result in over-smoothing or under-smoothing of images, particularly in overlap regions during step-and-shoot multi-bed imaging, leading to loss of clinically relevant features and noise issues due to axial truncation and limited field of view.
Innovation Solution
A method that computes a local penalty function for noise suppression, reducing its value in overlap regions by accounting for total coincidence counts and geometric sensitivity, allowing for dynamic adjustment of penalty parameters to preserve spatial resolution and avoid over- or under-smoothing, using a joint edge preservation parameter that combines data from adjacent volumes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a uniform penalty function is used across all volumes in step-and-shoot PET imaging, then the reconstruction process is simple, but over-smoothing occurs in overlap regions leading to loss of clinically relevant features
Solution Approach 1:
The patent applies local quality by computing separate penalty values for different volumes based on their specific characteristics. Each volume receives a tailored penalty function that accounts for its overlap status and count statistics, preventing uniform over-smoothing while maintaining computational feasibility. This resolves the contradiction by making the reconstruction process sufficiently simple while preserving critical image features through localized penalty adjustment.
2Object-affected harmful factors
If noise suppression is strengthened in overlap regions, then noise is reduced, but spatial resolution and clinically relevant features are lost
Solution Approach 1:
The patent dynamically changes the penalty parameter based on local conditions in each volume. By computing penalty values that reflect the actual count statistics and overlap characteristics of each volume, the method adapts the noise suppression strength locally. This prevents excessive smoothing in overlap regions while maintaining noise reduction where appropriate, thus preserving spatial resolution and clinically relevant features.
3Power
If separate reconstruction of each bed position is performed, then reconstruction computational load per volume is reduced, but noise in edge slices increases due to effective loss of data
Solution Approach 1:
The patent merges information from adjacent volumes by computing penalty functions that incorporate count statistics from overlapping regions. This allows the separate reconstruction approach to benefit from combined information, reducing noise in edge slices without requiring full joint reconstruction. The method achieves this by using the overlap data to inform penalty values, effectively combining information while maintaining computational efficiency.
4Productivity
If a fixed penalty parameter is used for all volumes, then the reconstruction process is computationally efficient, but the penalty is either too strong or too weak for specific volumes with different count statistics
Solution Approach 1:
The patent introduces dynamics by making the penalty parameter adaptive rather than fixed. The penalty value for each volume is computed dynamically based on its specific count statistics and overlap characteristics. This allows the reconstruction process to maintain computational efficiency while achieving accurate, volume-specific penalty application that adapts to varying data quality across different bed positions.
Data Source
AI summary
A non-transitory computer-readable medium stores instructions readable and executable by a workstation (18) including at least one electronic processor (20) to perform an imaging method (100). The method includes: receiving imaging data on a frame by frame basis for frames along an axial direction with neighboring frames overlapping along the axial direction wherein the frames include at least a volume (k) and a succeeding volume (k+1) at least partially overlapping the volume (k) along the axial direction; and generating an image of the volume (k) using an iterative image reconstruction process in which an iteration of the iterative image reconstruction process includes: computing a local penalty function for suppressing noise over the volume (k) including reducing the value of the local penalty function in an overlap region; generating an update image of the volume (k) using imaging data from the volume (k) and further using the local penalty function.