Virtual Frames for Distributed PET Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current PET imaging techniques, particularly with continuous bed movement (CBM), face challenges in reducing overall scan time, improving patient comfort, and managing intensive computing resources for image reconstruction, which leads to bottlenecks in clinical workflow and inefficient multi-modal image combination.
Innovation Solution
Implementing virtual frames for distributed list-mode reconstruction in PET systems, where coincident event pairs are categorized and reconstructed into spatially defined frames during continuous bed movement, allowing for concurrent and distributed processing, and integration with time-of-flight information for improved image reconstruction and axial sampling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If continuous bed movement (CBM) is used to reduce scan time, then overall data acquisition time is reduced, but image reconstruction is deferred until all data is acquired causing workflow bottlenecks
Solution Approach 1:
The patent divides the continuous data acquisition into discrete virtual frames that can be processed independently. Each virtual frame contains data from a specific spatial region and can be reconstructed separately, enabling parallel processing and eliminating the bottleneck of waiting for complete data acquisition before reconstruction begins.
Solution Approach 2:
The system performs preliminary categorization of coincident pairs into virtual frames during data acquisition. This preliminary organization allows reconstruction to begin on early frames while data collection continues, rather than waiting for all data to be collected before starting any reconstruction.
2Ease of operation
If step and shoot technique is used to generate elongated images, then region imaging is achieved, but the time to move and stop the bed prolongs overall scan time
Solution Approach 1:
The patent implements continuous bed movement to acquire data for elongated regions without the stop-and-go motion of step and shoot. The useful action of data acquisition continues uninterrupted while the bed moves through the imaging region, eliminating the time losses associated with repeated stopping and starting.
3Ease of operation
If continuous bed movement is used to eliminate stopping/starting motion, then patient comfort is improved and scan time is reduced, but a single large data set is collected requiring intensive computing resources
Solution Approach 1:
The large continuous data set is segmented into multiple virtual frames that can be processed in parallel. This segmentation reduces the memory and computing resource requirements at any given time, as reconstruction can proceed on individual frames rather than requiring all data to be held in memory simultaneously.
Solution Approach 2:
The system performs partial reconstruction on subsets of data (individual virtual frames) rather than waiting to process the complete data set. This allows reconstruction to proceed with partial data, reducing the immediate computing burden while maintaining overall image quality.
4Measurement precision
If data is binned into sinograms only after all data is collected, then uniform sampling is achieved, but reconstruction is deferred causing workflow bottlenecks
Solution Approach 1:
The patent segments the data binning process into virtual frame-level operations. Each virtual frame's data is binned into sinograms independently as it is acquired, rather than waiting for complete data collection. This maintains sampling uniformity within each frame while enabling staggered reconstruction.
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
This approach enhances patient comfort, streamlines clinical workflow, reduces scan latency, and enables efficient concurrent reconstruction, resulting in improved axial sensitivity and spatial resolution, while facilitating region-of-interest adapted acquisition and integrated multi-modal imaging.
Implementation Method 1
detector arrays detect pairs of gamma photon's emitted from a positron annihilation event in a subject
Implementation Method 2
A time-of-flight (TOF) PET adds an estimate of the originating location where the annihilation event occurred based on the mean time difference between detection of each photon pair
Data Source
AI summary
A positron emission tomography (PET) system includes a memory (18), a subject support (3), a categorizing unit (20), and a reconstruction unit (22). The memory (18) continuously records detected coincident event pairs detected by PET detectors (4). The subject support (3) supports a subject and moves in a continuous movement through a field of view (10) of the PET detectors (4). The categorizing unit (20) categorizes the recorded coincident pairs into each of a plurality of spatially defined virtual frame (14). The reconstruction unit (22) reconstructs the categorized coincident pairs of each virtual frame into a frame image and combines the frame images into a common elongated image.