CT Image Reconstruction Pre-fetching via Geometry Templates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The efficiency of Computed Tomography (CT) image reconstruction is limited by data bandwidth constraints during the loading and processing of projection data, necessitating an improved method for accessing and pre-fetching data to enhance the reconstruction process.
Innovation Solution
A method and system for determining and pre-fetching projection data involves calculating geometry to access and pre-fetch data sets from storage, using data templates to optimize data access and reduce overlapping regions, and utilizing multiple processors in a parallel architecture to efficiently generate images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If projection data is loaded and processed sequentially during CT image reconstruction, then data access is simple, but reconstruction efficiency is limited by data bandwidth constraints
Solution Approach 1:
The patent pre-calculates geometry information and pre-fetches projection data into a cache storage before the actual reconstruction process. By determining the sequence of pixels and calculating required projection data in advance, the system prepares data templates that guide efficient data retrieval, reducing the bandwidth burden during critical reconstruction operations
Solution Approach 2:
The patent divides the reconstruction process into discrete pixel sequences and segments projection data access into manageable chunks. By processing pixels in determined sequences and using data templates to segment the required projection data, the system optimizes cache utilization and reduces the overall data bandwidth requirements
2Speed
If all projection data is loaded into cache for processing, then data access speed improves, but cache memory requirements increase
Solution Approach 1:
The patent uses data templates to identify and load only the specific portions of projection data required for each pixel sequence into the cache. Rather than loading all projection data uniformly, the system selectively fetches local data segments based on the determined pixel sequence and geometry calculations, optimizing cache usage while maintaining fast access speeds
3Measurement precision
If geometry calculation is performed for each pixel independently, then processing accuracy is maintained, but processing time increases
Solution Approach 1:
The patent merges geometry calculations by determining a pixel sequence and calculating geometry information that serves multiple pixels in the sequence. By establishing data templates based on geometry calculations that cover ranges of pixels rather than individual pixels, the system maintains calculation accuracy while significantly reducing the time required for geometry processing
Data Source
AI summary
The disclosure relates to a system and method for determining and pre-fetching projection data in image reconstruction. The method may include: determining a sequence of a plurality of pixels including a first pixel and a second pixel relating to the first pixel; determining a first geometry calculation used for at least one processor to access a first set of projection data relating to the first pixel from a first storage; determining a second geometry calculation based on the first geometry calculation; determining a first data template relating to the first pixel and a second data template relating to the second pixel based on the second geometry calculation; and pre-fetching a second set of projection data based on the first data template and the second data template, from a storage.