Tomographic Motion Correction Using Projection-Based Motion Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing projection-based tomographic imaging techniques, such as CT and PET, suffer from reduced image quality due to object motion during data acquisition, particularly when using low doses of radioactive tracers, limiting the accuracy of imaging moving objects.
Innovation Solution
A system and method for motion-corrected imaging using acquired projection data and a motion estimation unit to determine a motion estimate based on a reference image and a motion model, allowing for the reconstruction of a more accurate motion-corrected image by associating the reference image with modelled projection data for candidate motions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If projection-based tomographic imaging is performed during an acquisition period, then image data can be obtained, but object motion during acquisition leads to degradation in image quality
Solution Approach 1:
The system performs preliminary motion estimation by comparing acquired projection data with modelled projection data generated from a reference image and motion model before final image reconstruction. This preliminary motion characterization allows the system to compensate for motion effects during the reconstruction process, thereby maintaining image quality despite the acquisition period required for data collection
Solution Approach 2:
The system employs an iterative feedback mechanism where motion estimates derived from comparing acquired and modelled projection data are fed back into the image reconstruction process. The motion-corrected images are then used to generate updated reference images for subsequent motion estimation cycles, continuously improving motion compensation and image quality throughout the acquisition period
2Measurement precision
If the acquisition period is extended to improve image quality, then more projection data can be acquired, but object motion during the extended period further degrades image quality
Solution Approach 1:
The system uses a dynamic motion model that can represent various types of object motion (periodic, non-periodic, rigid, non-rigid) rather than assuming a fixed motion pattern. The motion model parameters are continuously updated through iterative motion estimation, allowing the system to adapt to different motion scenarios and maintain correction accuracy even during extended acquisition periods
Solution Approach 2:
The system changes the parameters of the motion model during the acquisition period based on continuous motion estimation from acquired projection data. By updating motion model parameters iteratively and using these updated parameters for image reconstruction, the system maintains adaptability to changing motion patterns throughout the extended acquisition period
3Reliability
If existing motion correction approaches are used, then image quality can be maintained for specific kinds of motion, but accuracy is significantly limited for general motions
Solution Approach 1:
The system employs a universal motion model that can represent multiple types of motion (periodic, non-periodic, rigid, non-rigid) within a single unified framework. The motion estimation unit compares acquired projection data with modelled projection data generated from this versatile motion model, enabling reliable motion correction across diverse motion scenarios without requiring separate correction methods for different motion types
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to motion-corrected imaging of an object based on acquired projection data being tomographic projection data acquired for the object at a plurality of acquisition times during an acquisition period, and a motion estimate being indicative of a motion of the object in the acquisition period. The motion estimate is determined, based on the acquired projection data, a reference image of the object and a motion model indicating a plurality of candidate motions of the object during the acquisition period, by associating the reference image, for each of the plurality of candidate motions, with modelled projection data being modelled for the acquisition period assuming a respective candidate motion. A motion-corrected image of the object can then be reconstructed based on the acquired projection data and the motion estimate. In this way, an accurate imaging of moving objects can be achieved.