A Dynamic Pet Image Reconstruction Method Based on Tensor Dictionary Constraints

Abstract

The present invention discloses a dynamic PET image reconstruction method based on tensor dictionary constraint. The method is characterized by introducing a third-order tensor concept and the relevant tensor product definition to help the dynamic PET image reconstruction, establishing a tensor mathematical model of a reconstruction problem and adding the tensor dictionary constraint and the tensor dictionary-based constraint to carry out the dynamic PET image reconstruction, and finally adopting an alternating direction method of multipliers (ADMM) algorithm to optimize and solve, so that the tensor dictionary constraint is utilized effectively, and the problems of the result low resolution and the noise interference of a computer during a PET image reconstruction process are improved.

Description

technical field [0001] The invention belongs to the technical field of PET imaging, and in particular relates to a dynamic PET image reconstruction method based on tensor dictionary constraints. Background technique [0002] Positron Emission Tomography (PET) is a medical imaging technique based on nuclear physics and molecular biology. in accordance with. Dynamic PET scans the patient for a period of time to obtain data of many frames changing over time. This is a functional medical imaging mode that can record accurate pharmacokinetic quantitative information in vivo for early cancer detection and treatment. Response assessment provides effective assistance. [0003] The traditional dynamic PET reconstruction method is to perform static reconstruction on each frame of the dynamic data separately, so as to obtain the image set of the radioactivity concentration changing with time after reconstruction. PET images have the advantages of high sensitivity and high specificit...

AI Technical Summary

Problems solved by technology

PET images have the advantages of high sensitivity and high specificity, but due to the serious attenuation of radionuclides affected by human tissue, and the method of correcting the attenuation is complicated and costly, the image resolution reconstructed from the measured data low, the image is slightly blurred

Traditionally, statistical iterative methods are often used for the reconstruction of radioactive concentration distribution. Since the iterative method is based on statistical models and has good adaptability to incomplete data, it has gradually become the focus of research on PET reconstruction algorithms, including the famous ML-EM (Maximum Likelihood Expectation Maximization), MAP (maximum a posteriori) and SAGE (penalized likelihood) algorithms, however, these methods only consider the spatial information of each frame of data, ignoring the time connection between each frame of data

Images