Penalized PET Reconstruction Using Deep Learning Prior and Local Linear Fitting

Kim, Kyungsang; Wu, Dufan; Gong, Kuang; Dutta, Joyita; Kim, Jong Hoon; Son, Young-Don; Kim, Hang Keun; Fakhri, Georges El; Li, Quanzheng

doi:10.1109/tmi.2018.2832613

Cited by 180 publications

(110 citation statements)

References 26 publications

Supporting

Mentioning

110

Contrasting

Order By: Relevance

“…Another, more sophisticated strategy is to use the CNN output as a prior in reconstruction. We have previously used this approach and produced promising results for denoising [56] and anticipate that it will work for deblurring and SR by extension.…”

Section: Discussionmentioning

confidence: 99%

Super-Resolution PET Imaging Using Convolutional Neural Networks

Song

Chowdhury

Yang

et al. 2020

IEEE Trans. Comput. Imaging

Self Cite

View full text Add to dashboard Cite

Positron emission tomography (PET) suffers from severe resolution limitations which limit its quantitative accuracy. In this paper, we present a super-resolution (SR) imaging technique for PET based on convolutional neural networks (CNNs). To facilitate the resolution recovery process, we incorporate high-resolution (HR) anatomical information based on magnetic resonance (MR) imaging. We introduce the spatial location information of the input image patches as additional CNN inputs to accommodate the spatially-variant nature of the blur kernels in PET. We compared the performance of shallow (3-layer) and very deep (20-layer) CNNs with various combinations of the following inputs: low-resolution (LR) PET, radial locations, axial locations, and HR MR. To validate the CNN architectures, we performed both realistic simulation studies using the BrainWeb digital phantom and clinical neuroimaging data analysis. For both simulation and clinical studies, the LR PET images were based on the Siemens HR+ scanner. Two different scenarios were examined in simulation: one where the target HR image is the ground-truth phantom image and another where the target HR image is based on the Siemens HRRT scanner -a high-resolution dedicated brain PET scanner. The latter scenario was also examined using clinical neuroimaging datasets. A number of factors affected relative performance of the different CNN designs examined, including network depth, target image quality, and the resemblance between the target and anatomical images. In general, however, all CNNs outperformed classical penalized deconvolution techniques by large margins both qualitatively (e.g., edge and contrast recovery) and quantitatively (as indicated by two metrics: peak signal-to-noise-ratio and structural similarity index).

show abstract

Section: Discussionmentioning

confidence: 99%

Super-Resolution PET Imaging Using Convolutional Neural Networks

Song

Chowdhury

Yang

et al. 2020

IEEE Trans. Comput. Imaging

Self Cite

View full text Add to dashboard Cite

show abstract

“…In this work, a template image and a tissue kinetic model were used to simulate PET sinogram data and a thinning Poisson process was used to generate the dynamic PET data. The thinning Poisson process has been used by others for the evaluation of PET image reconstruction methods (Kim et al 2018). However, the simulated data would be more realistic if a full Monte Carlo simulation of positron emission decay and the PET detection process at the coincidence event level (e.g.…”

Section: Discussionmentioning

confidence: 99%

Analysis of continuous infusion functional PET (fPET) in the human brain

Jamadar

Ward

et al. 2019

Preprint

View full text Add to dashboard Cite

Functional positron emission tomography (fPET) is a neuroimaging method involving continuous infusion of 18-F-fluorodeoxyglucose (FDG) radiotracer during the course of the PET examination. Compared with the conventional bolus administered static FDG PET which provides only a snapshot of the averaged glucose uptake into the brain in a limited dynamic time window, fPET offers a significantly wider time window to study the dynamics of glucose uptake. Several earlier studies have applied fPET to investigate brain FDG uptake and study its relationship with functional magnetic resonance imaging (fMRI). However, due to the unique characteristics of fPET signals, modelling of the fPET signal is a complex task and poses challenges for accurate interpretation of the results. This study applies independent component analysis (ICA) to analyze resting state fPET data, and to compare the performance of ICA and general linear modelling (GLM) for estimation of brain activation in response to tasks. The fPET signal characteristics were compared using GLM and ICA methods to model the fPET visual activation data. Our aim was to evaluate GLM and ICA methods for analyzing task fPET datasets and present ICA method in the analysis of resting state fPET datasets. Using both simulation and in-vivo experimental datasets, we show that both methods can successfully identify task related brain activation. We report fPET metabolic resting state brain networks analyzed using the fPET ICA method in a cohort of healthy subjects. Functional PET provides a unique method to map dynamic changes of glucose uptake in the resting human brain and in response to extrinsic stimulation.

show abstract

“…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).…”

Section: Introductionmentioning

confidence: 99%

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

Lim

Dewaraja

Fessler

2020

Medical Imaging 2020: Physics of Medical Imaging

View full text Add to dashboard Cite

Image reconstruction in low-count PET is particularly challenging because gammas from natural radioactivity in Lu-based crystals cause high random fractions that lower the measurement signal-to-noise-ratio (SNR). In model-based image reconstruction (MBIR), using more iterations of an unregularized method may increase the noise, so incorporating regularization into the image reconstruction is desirable to control the noise. New regularization methods based on learned convolutional operators are emerging in MBIR. We modify the architecture of a variational neural network, BCD-Net, for PET MBIR, and demonstrate the efficacy of the trained BCD-Net using XCAT phantom data that simulates the low true coincidence countrates with high random fractions typical for Y-90 PET patient imaging after Y-90 microsphere radioembolization. Numerical results show that the proposed BCD-Net significantly improves PET reconstruction performance compared to MBIR methods using non-trained regularizers, total variation (TV) and non-local means (NLM), and a non-MBIR method using a single forward pass deep neural network, U-Net. BCD-Net improved activity recovery for a hot sphere significantly and reduced noise, whereas non-trained regularizers had a trade-off between noise and quantification. BCD-Net improved CNR and RMSE by 43.4% (85.7%) and 12.9% (29.1%) compared to TV (NLM) regularized MBIR. Moreover, whereas the image reconstruction results show that the non-MBIR U-Net over-fits the training data, BCD-Net successfully generalizes to data that differs from training data. Improvements were also demonstrated for the clinically relevant phantom measurement data where we used training and testing datasets having very different activity distribution and countlevel.Index Terms-Variational neural network, Regularized modelbased image reconstruction, Low-count quantitative PET, Y-90This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible.

show abstract

Penalized PET Reconstruction Using Deep Learning Prior and Local Linear Fitting

Cited by 180 publications

References 26 publications

Super-Resolution PET Imaging Using Convolutional Neural Networks

Super-Resolution PET Imaging Using Convolutional Neural Networks

Analysis of continuous infusion functional PET (fPET) in the human brain

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

Contact Info

Product

Resources

About