MLAA and DCC Algorithms for PET Attenuation Correction Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current nuclear medical imaging techniques, particularly PET, face challenges in generating accurate attenuation correction maps due to truncation issues in integrated PET-CT and PET-MR systems, especially with non-uniform biodistribution tracers and the presence of metal implants, leading to image artifacts and noise.
Innovation Solution
The MLAA algorithm is modified and expanded to handle non-uniform biodistribution tracers, combined with the DCC algorithm to generate accurate mu-maps, and used in gated and dynamic MR/PET cardiac studies to reduce noise and artifacts, while segmenting and replacing truncated regions with MR-based data for improved attenuation correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional attenuation correction methods are used in PET imaging, then the imaging process is simple and fast, but image accuracy deteriorates due to truncation artifacts and noise, especially with non-uniform biodistribution tracers and metal implants
Solution Approach 1:
The patent segments the attenuation correction process into multiple components: truncated region identification, MLAA-based estimation for truncated regions, and DCC-based refinement. This segmentation allows each component to address specific aspects of the problem, improving overall accuracy while managing complexity through modular processing
Solution Approach 2:
The patent performs preliminary identification and segmentation of truncated regions before applying correction algorithms. By pre-processing the data to identify problem areas, the subsequent MLAA and DCC algorithms can focus computational resources on correcting specific regions, improving efficiency and accuracy
2Measurement precision
If MLAA algorithm is applied to estimate missing mu-values in truncated regions, then attenuation correction accuracy is improved, but computational time and complexity increase
Solution Approach 1:
The patent applies MLAA selectively only to truncated regions rather than the entire image, and uses a limited number of iterations (e.g., 5-10 iterations). This partial application reduces computational burden while maintaining accuracy in the critical truncated areas where it is most needed
Solution Approach 2:
The patent introduces DCC as an intermediary refinement step that uses the MLAA-estimated mu-map as input. The DCC algorithm performs a consistency check and refinement, improving accuracy while requiring less computation than a full MLAA execution, thus balancing precision and processing time
3Manufacturing precision
If DCC algorithm is used to refine mu-map estimation, then image reconstruction accuracy is improved, but the processing complexity and time increase
Solution Approach 1:
The patent implements a feedback mechanism where DCC uses the MLAA-generated mu-map as input and refines it by checking data consistency conditions. The refined mu-map then feeds into the final image reconstruction process. This feedback loop ensures accuracy while maintaining manageable complexity through iterative refinement rather than complex single-step processing
4Reliability
If truncated regions are replaced with MR-based data, then artifacts are reduced, but the integration of multiple imaging modalities increases system complexity
Solution Approach 1:
The patent merges data from PET and MR imaging modalities by using MR images to identify truncated regions and providing anatomical context for attenuation correction. This merging leverages the complementary strengths of both modalities: PET for functional information and MR for soft tissue contrast and anatomical boundaries, improving reliability while managing integration complexity through standardized processing pipelines
Data Source
AI summary
The DCC (Data Consistency Condition) algorithm is used in combination with MLAA (Maximum Likelihood reconstruction of Attenuation and Activity) to generate extended attenuation correction maps for nuclear medicine imaging studies. MLAA and DCC are complementary algorithms that can be used to determine the accuracy of the mu-map based on PET data. MLAA helps to estimate the mu-values based on the biodistribution of the tracer while DCC checks if the consistency conditions are met for a given mu-map. These methods are combined to get a better estimation of the mu-values. In gated MR/PET cardiac studies, the PET data is framed into multiple gates and a series of MR based mu-maps corresponding to each gate is generated. The PET data from all gates is combined. Once the extended mu-map is generated the central region is replaced with the MR based mu-map corresponding to that particular gate. On the other hand, in dynamic PET studies the uptake in the patient's arms reaches a steady state only after the tracer distributes throughout the body. Hence, for dynamic scans, the projection data of all frames is summed and used to generate the MLAA based extended mu-map for all frames.