Iterative CT Reconstruction for Spect/CT Truncation Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reconstruction methods in nuclear and x-ray computed tomography imaging face challenges with field-of-view truncation errors, leading to artifacts and inaccurate attenuation correction, especially in SPECT and PET imaging, where photon absorption and scatter distort images, and iterative methods are prone to truncation errors.
Innovation Solution
The method involves acquiring projection data and extending the field-of-view (FoV) to encompass the subject's maximum trans-axial extents, using a voxel grid to iteratively reconstruct the data, thereby reducing truncation errors and providing accurate attenuation and scatter corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative reconstruction methods are used to improve image quality and accommodate corrections for attenuation and noise, then image accuracy is improved, but truncation errors occur leading to artifacts when the reconstruction FoV does not encompass the whole imaged object
Solution Approach 1:
The system performs preliminary detection of the subject's maximum trans-axial extents before performing the iterative reconstruction. This preliminary action allows the reconstruction FoV to be properly configured to encompass the entire object, preventing truncation errors from occurring during the reconstruction process while still maintaining the benefits of iterative methods for accurate attenuation correction and noise handling.
2Reliability
If the reconstruction FoV is increased to encompass the whole imaged object to remove truncation artifacts, then reconstruction reliability is improved, but the complexity of determining the appropriate FoV increases
Solution Approach 1:
The system performs preliminary detection of the subject's maximum trans-axial extents using the acquired projection data before reconstruction. This automated preliminary measurement simplifies the process of determining the appropriate FoV by objectively identifying the subject boundaries, eliminating the need for complex manual FoV configuration while ensuring the entire object is encompassed for reliable reconstruction.
Solution Approach 2:
The system uses the acquired projection data itself to automatically determine the subject's maximum trans-axial extents and configure the reconstruction FoV accordingly. This self-service approach allows the reconstruction process to automatically adapt to the subject size and shape without requiring external input or complex preprocessing, simplifying the overall workflow while ensuring reliable reconstruction coverage.
3Productivity
If FBP reconstruction is used to achieve computational speed and ease of implementation, then processing efficiency is improved, but image noise increases due to inability to model low photon counts
Solution Approach 1:
The system applies iterative reconstruction methods which, while more computationally intensive than FBP, provide the necessary capability to model low photon counts and reduce image noise. The use of extended FoV with automated extent detection enables this more thorough reconstruction approach to be applied efficiently by eliminating truncation artifacts that would otherwise require additional processing steps, thus achieving better SNR without excessive computational overhead.
Data Source
Figure 1
Figure 2A~2C
Figure 3
AI summary
A multiple modality imaging system (10) includes a cone-beam computed tomography scanner (24, 30) which acquires CT projection data of a subject (22) in an examination region (18) and a nuclear imaging scanner which concurrently/subsequently acquires nuclear projection data of the subject in the examination region. A CT reconstruction processor (34) is programmed to perform the steps of: in the CT projection data, defining a field-of-view (FoV) with a voxel grid in a trans-axial direction; determining the subject's maximum trans-axial extents; generating an extending FoV by extending the voxel grid of the FoV to at least one extended region outside the FoV that encompass at least the determined maximum trans-axial extents and all attenuation in the trans-axial direction; and iteratively reconstructing the CT projection data into an attenuation map of the extended FoV. At least one nuclear reconstruction processor (44) is programmed to correct the acquired nuclear projection data based on the iteratively reconstructed attenuation map.