Abstract:Our aim was to enhance visual quality and quantitative accuracy of dynamic positron emission tomography (PET)uptake images by improved image reconstruction, using sophisticated sparse penalty models that incorporate both 2D spatial+1D temporal (3DT) information. We developed two new 3DT PET reconstruction algorithms, incorporating different temporal and spatial penalties based on discrete cosine transform (DCT)w/ patches, and tensor nuclear norm (TNN) w/ patches, and compared to frame-by-frame methods; conventional 2D ordered subsets expectation maximization (OSEM) w/ post-filtering and 2D-DCT and 2D-TNN. A 3DT brain phantom with kinetic uptake (2-tissue model), and a moving 3DT cardiac/lung phantom was simulated and reconstructed. For the cardiac/lung phantom, an additional cardiac gated 2D-OSEM set was reconstructed. The structural similarity index (SSIM) and relative root mean squared error (rRMSE) relative ground truth was investigated. The image derived left ventricular (LV) volume for the cardiac/lung images was found by region growing and parametric images of the brain phantom were calculated. For the cardiac/lung phantom, 3DT-TNN yielded optimal images, and 3DT-DCT was best for the brain phantom. The optimal LV volume from the 3DT-TNN images was on average 11 and 55 percentage points closer to the true value compared to cardiac gated 2D-OSEM and 2D-OSEM respectively. Compared to 2D-OSEM, parametric images based on 3DT-DCT images generally had smaller bias and higher SSIM. Our novel methods that incorporate both 2D spatial and 1D temporal penalties produced dynamic PET images of higher quality than conventional 2D methods, w/o need for post-filtering. Breathing and cardiac motion were simultaneously captured w/o need for respiratory or cardiac gating. LV volumes were better recovered, and subsequently fitted parametric images were generally less biased and of higher quality.

3D Tensor Based Nonlocal Low Rank Approximation in Dynamic PET Reconstruction.

Nonlocal Total Variation Based Dynamic PET Image Reconstruction with Low-Rank Constraints

Nonlocal Low-Rank and Total Variation Constrained PET Image Reconstruction

Improved Robust Tensor Principal Component Analysis for Accelerating Dynamic MR Imaging Reconstruction

Simultaneous Reconstruction and Segmentation of Dynamic Pet Via Low-Rank and Sparse Matrix Decomposition

Sparse/Low Rank Constrained Reconstruction for Dynamic PET Imaging.

Dynamic Low-Count PET Image Reconstruction Using Spatio-Temporal Primal Dual Network.

Deep Reconstruction Model For Dynamic Pet Images

Improved Patlak Reconstruction from Low-dose Dynamic PET Using Temporal Non-local Neural Network

Simultaneous Reconstruction and Segmentation for Dynamic PET: A Low Rank Framework.

Joint Reconstruction of Dynamic PET Activity and Kinetic Parametric Images Using Total Variation Constrained Dictionary Sparse Coding

Deep Dual-Domain Based Framework for PET Image Reconstruction

Low dose PET reconstruction with total variation regularization.

Dynamic PET cardiac and parametric image reconstruction: a fixed-point proximity gradient approach using patch-based DCT and tensor SVD regularization

Rapid high-quality PET Patlak parametric image generation based on direct reconstruction and temporal nonlocal neural network.

PEt image reconstruction based on particle filter framework

Uncertainty Penalized Weighted Least Squares Framework For Pet Reconstruction Under Uncertain System Models

Structure and Tracer Kinetics-Driven Dynamic PET Reconstruction

Rethinking PET Image Reconstruction: Ultra-Low-Dose, Sinogram and Deep Learning

Robust Framework for PET Image Reconstruction Incorporating System and Measurement Uncertainties.

Low Dose PET Image Reconstruction with Total Variation Using Alternating Direction Method.