Iterative PET Image Reconstruction Using Convolutional Neural Network Representation

Abstract

PET image reconstruction is challenging due to the ill-poseness of the inverse problem and limited number of detected photons. Recently deep neural networks have been widely and successfully used in computer vision tasks and attracted growing interests in medical imaging. In this work, we trained a deep residual convolutional neural network to improve PET image quality by using the existing inter-patient information. An innovative feature of the proposed method is that we embed the neural network in the iterative reconstruction framework for image representation, rather than using it as a post-processing tool. We formulate the objective function as a constrained optimization problem and solve it using the alternating direction method of multipliers (ADMM) algorithm. Both simulation data and hybrid real data are used to evaluate the proposed method. Quantification results show that our proposed iterative neural network method can outperform the neural network denoising and conventional penalized maximum likelihood methods.

Joint low-count PET/CT segmentation and reconstruction with paired variational neural networks

“…We implemented non-recurrent (single forward pass) 3-D version of U-Net to compare with the proposed BCD-Net. The 'encoder' part of U-Net consists of multiple sets of 1) max pooling layer, 2) 3×3×3 convolutional layer, 3) batch normalization (BN) layer, 4) ReLU layer and the 'decoder' part of U-Net consists of multiple sets of 1) upsampling with trilinear interpolation [17], 2) 3×3×3 convolutional layer, 3) batch normalization (BN) layer, 4) ReLU layer. We used ReLU layer as the last step to enforce the non-negativity constraint on image [17].…”

“…This recurrent framework enables NNs in the later stages to learn how to recover fine details. Our proposed BCD-Net is also distinct from [17], [18] in that denoising NNs are derived by variational (optimization) formulation with a mathematical motivation (whereas, for the trained regularizer, [17], [18] brought U-Net [19] and DnCNN [20] developed for medical image segmentation and general Gaussian denoising) and characterized by less parameters, thereby avoiding overfitting and generalizing well to unseen data especially when training samples are limited (see Section IV).…”

Image Reconstruction: From Sparsity to Data-Adaptive Methods and Machine Learning

“…2) Hybrid-domain learning: In this class of approaches [33], [142], [142], [159], [160], [164]- [166], [184]- [189], the data consistency term is imposed in the neural network training and inference to improve the performance as shown in Fig. 7(b).…”

“…Specifically, in CNN penalty approaches [164], a neural network is used as a prior model within an MBIR framework. Rather than using a CNN penalty explicitly, in the plug-and-play approach [188], [189], the denoising step of an iteration like ADMM is replaced with a neural network denoiser. Similarly, the deep image prior approach [192] formulates the image reconstruction problem as…”

PET/MRI attenuation estimation in the lung: A review of past, present, and potential techniques

“…Oxygen-enhanced MRI can be used to shorten lung tissue, blood, and plasma T1 values, potentially giving ventilation and perfusion MRI weighting, 31 although lung registration between pre and postinhalation datasets is a challenge. 44 Signal can be provided from hyperpolarized 129 Xe or 3 He gas or inhaled inert fluorinated ( 19 F) gas. 44…”