Iterative Regularized Reconstruction for PET Image Artifact Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for quantifying nuclear imaging and tomographic brain images require invasive blood sampling, which is costly, risky, and limits clinical utility, and iterative reconstruction with point spread function (PSF) modeling introduces ringing artifacts and over-enhancement that need to be mitigated.
Innovation Solution
The development of iterative regularized reconstruction methods that incorporate locally-weighted total variation denoising to suppress artifacts induced by PSF modeling, allowing for zero-blood-sample imaging and maintaining contrast recovery, using a computer-implemented method that calculates weights for each voxel based on convergence rate and performs TV-gPSF-MLEM reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative reconstruction with point spread function (PSF) modeling is used to improve contrast recovery, then contrast recovery is improved, but ringing artifacts and over-enhancement are introduced
Solution Approach 1:
The patent introduces a regularization term as an intermediary element between the PSF modeling and the final image reconstruction. This regularization term acts as a mediator that suppresses the harmful ringing artifacts while preserving the beneficial contrast recovery effects of PSF modeling. The regularization parameter controls the balance between artifact suppression and contrast enhancement.
Solution Approach 2:
The patent modifies the reconstruction parameters by introducing a regularization parameter that dynamically adjusts the weighting between fidelity to the measured data and suppression of artifacts. By changing this parameter, the system can control the degree of artifact suppression while maintaining contrast recovery, effectively resolving the contradiction between improving contrast and reducing ringing artifacts.
2Measurement precision
If blood sampling is performed to measure tracer blood concentration, then quantification accuracy is improved, but patient risk and cost increase
Solution Approach 1:
The patent enables the imaging system to self-determine the tracer blood concentration curve by extracting this information directly from the PET image data and kinetic modeling. Instead of requiring external blood sampling to obtain this information, the system uses its own imaging capabilities and computational methods to derive the necessary physiological parameters, thereby eliminating the harmful effects of invasive blood sampling while maintaining quantification accuracy.
3Manufacturing precision
If iterative reconstruction with PSF modeling is used to improve image quality, then small structure detection is improved, but noise amplification occurs
Solution Approach 1:
The regularization term serves as an intermediary that selectively suppresses noise amplification while preserving the enhanced small structure detection capabilities provided by PSF modeling. The regularization acts as a filter that distinguishes between meaningful small structure signals and spurious noise, allowing the system to maintain high detection precision without suffering from excessive noise amplification.
Data Source
AI summary
The present disclosure is directed to iterative regularized reconstruction methods. In certain embodiments, the methods incorporate locally-weighted total variation denoising to suppress artifacts induced by PSF modeling. In certain embodiments, the methods are useful for suppressing ringing artifacts while contrast recovery is maintained. In certain embodiments, the weighting scheme can be extended to noisy measures introducing a noise-independent weighting scheme. The present disclosure is also directed to a method for quantifying radioligand binding in a subject without collecting arterial blood. In certain embodiments, the methods incorporate using imaging data and electronic health records to predict one or more anchors, which are used to generate an aterial input function (AIF) for the radioligand.