Dynamic PET Imaging Neural Network Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional PET imaging methods struggle with fast reconstruction of dynamic images, especially in long scans, as they require processing multiple frames to account for motion and tracer kinetics, which can lead to delayed image availability.
Innovation Solution
A system and method utilizing a trained neural network to back-project PET datasets into histo-image frames, allowing for real-time generation of dynamic PET images by processing time-referenced data, enabling simultaneous monitoring of changes during the scan.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional PET reconstruction methods are used to process multiple frames for dynamic imaging, then motion correction and tracer kinetics assessment are improved, but image reconstruction time increases significantly
Solution Approach 1:
The system performs preliminary back-projection of the entire dynamic PET dataset into a sequence of histo-image frames before final image reconstruction. This preliminary organization of data into time-resolved frames enables faster subsequent reconstruction while preserving motion correction capabilities, as the frames are pre-prepared and can be processed more efficiently by the reconstruction algorithm.
Solution Approach 2:
The patent segments the dynamic PET dataset into multiple time-resolved histo-image frames through back-projection, where each frame corresponds to a specific time point. This segmentation allows the reconstruction process to work with smaller, time-specific subsets of data rather than processing the entire dynamic dataset as one large block, thereby reducing overall reconstruction time while maintaining the ability to correct for motion across different time points.
2Measurement precision
If multiple time frames are reconstructed for dynamic PET imaging, then tracer kinetics analysis is improved, but computational complexity increases
Solution Approach 1:
The system performs preliminary back-projection of the dynamic PET dataset into time-resolved histo-image frames before final reconstruction. This preliminary step organizes the data into a structured format where each frame corresponds to a specific time point, enabling more efficient subsequent processing and reducing the computational burden of reconstructing multiple frames while preserving tracer kinetics information.
Solution Approach 2:
The patent creates multiple copies of the back-projection operation, each generating a time-resolved histo-image frame. Instead of performing a single complex reconstruction on the entire dynamic dataset, the system creates simplified copies (frames) that represent different time points, which can then be processed more efficiently while collectively providing the full dynamic information needed for tracer kinetics analysis.
3Productivity
If fast reconstruction algorithms are used for dynamic PET imaging, then image availability time is improved, but image quality may deteriorate
Solution Approach 1:
The system performs preliminary back-projection into time-resolved histo-image frames, which organizes the data in a way that enables faster subsequent reconstruction without sacrificing quality. This preliminary step creates a structured representation of the dynamic data that can be processed more efficiently while maintaining the fidelity needed for high-quality image reconstruction.
Solution Approach 2:
By segmenting the dynamic PET data into multiple time-resolved histo-image frames, the system enables parallel or sequential processing of smaller data subsets. This segmentation allows fast reconstruction algorithms to work on manageable frame-sized datasets rather than the entire dynamic dataset, achieving both speed and quality by processing divided portions that can be recombined into the final high-quality dynamic images.
Data Source
AI summary
Systems and methods of dynamic PET imaging are disclosed. A system includes a positron emission tomography (PET) imaging modality configured to execute a first scan to acquire a first PET dataset and a processor. The first PET dataset includes dynamic PET data. The processor is configured to back-project the first PET dataset to generate a plurality of histo-image frames, input each of the plurality of histo-image frames to a trained neural network, and receive a dynamic PET output from the trained neural network. Each of the histo-image frames corresponds to a first axial position of the PET imaging modality.