ABSTRACT
CONTENTS 5.1 Introduction 64 5.2 From 3D Method to 4D Method 64
5.2.1TV Regularization 65 5.2.2Solution Algorithm via ADMM 66 5.2.3Parallelization via ADMM 68 5.2.4Numerical Results 70
5.3Advanced Temporal Regularization Methods 71 5.3.1High-Order Local Methods 71 5.3.2Nonlocal Methods 75 5.3.3Low-Rank Models 76 5.3.4Learning-Based Algorithms 78 5.3.5 Others 79
5.4 Conclusion 79 References 80
In 4DCT [1,2] or 4DCBCT [3,4], a temporal sequence of CT images are reconstructed instead of a single static image, for which a respiratory signal is oen recorded to assist the binning of projection data to their belonging temporal frames, and then the images can be reconstructed either independently frame by frame (so-called 3D method) or altogether using temporal regularization methods (so-called 4D method).
