PET Attenuation Correction via Reversible Flow Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for attenuation correction in PET imaging require additional CT or MR scans, which increase costs, time, and radiation exposure, while also leading to inaccurate results, especially in lungs.
Innovation Solution
A PET system attenuation correction method based on a flow model that eliminates the need for additional scans by using a fully reversible flow model to directly convert non-attenuation-corrected PET images to attenuation-corrected images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If additional CT or MR scans are performed for attenuation correction, then attenuation correction accuracy is improved, but examination cost and time increase
Solution Approach 1:
The patent creates a synthetic attenuation map by copying and transforming information from the NAC PET image through a trained flow model, replacing the need for separate CT or MR scans. The flow model learns the mapping from NAC PET to attenuation map, effectively copying the necessary attenuation information from the already-acquired PET data.
Solution Approach 2:
The NAC PET image serves multiple functions: it is both the input for diagnosis and the source information for generating the attenuation map. The flow model enables one image to fulfill multiple roles, eliminating the need for separate attenuation correction scans and reducing overall examination time.
2Measurement precision
If additional CT scans are performed for attenuation correction, then attenuation correction accuracy is improved, but radiation exposure increases
Solution Approach 1:
Instead of acquiring additional CT data that would expose patients to more radiation, the patent copies the necessary attenuation information from the NAC PET image through the flow model. This approach obtains accurate attenuation correction without the harmful radiation of additional CT scans.
Solution Approach 2:
The patent converts the limitation of NAC PET images (which lack accurate attenuation information) into a benefit by using the flow model to generate the attenuation map from the available NAC data, avoiding the need for harmful additional CT radiation while still achieving accurate attenuation correction.
3Adaptability or versatility
If deep learning methods are used for attenuation correction without consistency constraints, then training flexibility is improved, but overfitting occurs and reliability decreases
Solution Approach 1:
The flow model incorporates a consistency constraint that provides feedback during training: the model must ensure that the generated attenuation map, when used to correct the NAC PET image, reproduces the original NAC image. This feedback mechanism prevents overfitting while maintaining training flexibility.
Solution Approach 2:
The flow model uses an invertible architecture where the forward transformation (NAC PET to attenuation map) has a corresponding inverse transformation. This inversion property enforces consistency constraints naturally, ensuring that the model learns reliable mappings without overfitting, while still allowing flexible training.
Data Source
AI summary
The present invention discloses a PET system attenuation correction method based on a flow model. The flow model adopted is a completely reversible model, the forward and reverse mappings share the same parameters, and the structure itself is a consistency constraint. The model utilizes the spatial correlation of adjacent slices, and adopts the structure of multi-slice input and single-slice output. The model consists of multiple reversible blocks, each of which consists of a enhanced affine coupling layer and a reversible 1×1 convolutional layer, and uses several small u-nets to learn the transformation parameters of the enhanced affine coupling layer. The present invention avoids additional CT or MR scanning, saves scanning cost for the patient, and reduces the damage of CT radiation to the patient; compared with similar methods, higher quality non-attenuation-corrected PET image can be obtained.