PET List-Mode Image Reconstruction With Multi-Level Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
List-mode reconstruction in PET imaging is too slow for clinical use due to the large number of events, and existing methods fail to provide consistent reconstruction time and image quality across varying scan durations.
Innovation Solution
A multi-level iterative reconstruction approach that divides coincidence events into representative groups, using different numbers of events per update to ensure consistent reconstruction time and quality, while preserving temporal information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If list-mode reconstruction is used to preserve temporal information, then image quality and motion correction capability are improved, but reconstruction speed becomes too slow for clinical settings
Solution Approach 1:
The patent segments the large set of coincidence events into multiple smaller groups, where each group contains a representative subset of events. This segmentation allows the reconstruction algorithm to process smaller data portions iteratively, significantly reducing computation time per iteration while maintaining the benefits of list-mode reconstruction for temporal information preservation and motion correction.
Solution Approach 2:
The patent applies partial action by using a representative subset of coincidence events in each reconstruction iteration rather than processing all events. Each group contains enough events to maintain image quality and temporal information, but processes only a portion of the total data per iteration, achieving faster overall reconstruction speed suitable for clinical settings.
2Productivity
If traditional sinogram reconstruction is used to increase reconstruction speed, then productivity is improved, but temporal information is lost and motion correction becomes unsatisfactory
Solution Approach 1:
The patent segments the coincidence event data into multiple representative groups, allowing iterative processing that preserves temporal information. By processing segmented groups rather than aggregating all data into sinograms, the method maintains event-by-event temporal details while achieving faster reconstruction through efficient iterative updates.
Solution Approach 2:
The patent uses partial action by processing representative subsets of events in each iteration rather than using all events simultaneously as in traditional sinogram reconstruction. This approach preserves temporal information from individual coincidence events while achieving reconstruction speeds suitable for clinical use through efficient iterative processing.
3Manufacturing precision
If all coincidence events are processed in each iteration to maintain image quality, then manufacturing precision is improved, but reconstruction time increases significantly
Solution Approach 1:
The patent segments the complete set of coincidence events into multiple representative groups that are processed across multiple iterations. Each group contains enough events to maintain image quality, and the iterative processing of segmented groups reduces the computational burden per iteration, significantly decreasing total reconstruction time while preserving image quality.
Solution Approach 2:
The patent applies partial action by processing representative subsets of coincidence events in each iteration rather than all events. Each subset contains sufficient events to maintain manufacturing precision and image quality, but processing only a portion of the total data per iteration dramatically reduces reconstruction time for clinical applicability.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Reduces reconstruction time by half or more compared to traditional methods while maintaining image quality comparable to maximum likelihood expectation maximization, suitable for clinical settings.
Implementation Method 1
each detector includes one or more scintillation crystals and one or more photosensors
Implementation Method 2
each detector includes one or more scintillation crystals and one or more photosensors
Implementation Method 3
A coincidence event occurs when a positron emitted by radionuclide collides with an electron and a pair of photons are emitted due to collision and annihilation of the positron and the electron
Data Source
AI summary
A positron emission tomography (PET) system is provided. The system includes an image reconstruction computing device. The processors of the image reconstruction computing device are programmed to receive event data acquired by the PET system. The event data are represented as a list of coincidence events. The processors are also programmed to generate groups of coincidence events based on the event data, each group being representative of the event data. The processors are further programmed to perform a first level of image updates by iteratively updating a reconstructed image. Each image update is based on a first number of coincidence events. Further, the processors are programmed to perform a second level of image updates by iteratively updating the reconstructed image. Each image update is based on a second number of coincidence events. The first number of coincidence events is different from the second number of coincidence events.