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

Although Positron emission tomography (PET) has a wide range of clinical applications, radiation exposure to patients in PET continues to draw concerns. To reduce the radiation risk, efforts have been made to obtain high resolution images from low-resolution images. However, previous studies mainly focused on denoising PET images in image space, which ignored the influence of sinogram quality and constraints in reconstruction process. This paper proposed a directly reconstruction framework from ultra-low-dose sinogram based on deep learning. Two coupled networks are introduced to sequentially denoise low-dose sinogram and reconstruct the activity map. Evaluation on in vivo PET dataset indicates that the proposed method can achieve better performance than other state-of-the-art methods and reconstruct satisfactory PET images with only 0.2% dose of standard one.

[1]  Lei Zhang,et al.  Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising , 2016, IEEE Transactions on Image Processing.

[2]  L. Shepp,et al.  Maximum Likelihood Reconstruction for Emission Tomography , 1983, IEEE Transactions on Medical Imaging.

[3]  R. Brooks,et al.  Statistical limitations in x-ray reconstructive tomography. , 1976, Medical physics.

[4]  Alexei A Samsonov,et al.  Noise‐adaptive nonlinear diffusion filtering of MR images with spatially varying noise levels , 2004, Magnetic resonance in medicine.

[5]  Aaron C. Courville,et al.  Improved Training of Wasserstein GANs , 2017, NIPS.

[6]  Thomas J. Fuchs,et al.  DeepPET: A deep encoder–decoder network for directly solving the PET image reconstruction inverse problem , 2018, Medical Image Anal..

[7]  Alexei A. Efros,et al.  Image-to-Image Translation with Conditional Adversarial Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Huafeng Liu,et al.  Three-dimensional convolutional neural networks for simultaneous dual-tracer PET imaging , 2019, Physics in medicine and biology.

[9]  M. Rafecas,et al.  PET Reconstruction From Truncated Projections Using Total-Variation Regularization for Hadron Therapy Monitoring , 2013, IEEE Transactions on Nuclear Science.

[10]  Harri Valpola,et al.  Weight-averaged consistency targets improve semi-supervised deep learning results , 2017, ArXiv.

[11]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[12]  Jiahe Tian,et al.  PET image denoising using unsupervised deep learning , 2019, European Journal of Nuclear Medicine and Molecular Imaging.

[13]  John M Pauly,et al.  Ultra-Low-Dose 18F-Florbetaben Amyloid PET Imaging Using Deep Learning with Multi-Contrast MRI Inputs. , 2019, Radiology.

[14]  Jens Gregor,et al.  CNN-based PET sinogram repair to mitigate defective block detectors , 2019, Physics in medicine and biology.

[15]  Gengsheng L. Zeng.  Image Noise Covariance Can be Adjusted by a Noise Weighted Filtered Backprojection Algorithm , 2019, IEEE Transactions on Radiation and Plasma Medical Sciences.

[16]  Tie-Yan Liu,et al.  Learning to rank: from pairwise approach to listwise approach , 2007, ICML '07.

[17]  Dinggang Shen,et al.  Semisupervised Tripled Dictionary Learning for Standard-Dose PET Image Prediction Using Low-Dose PET and Multimodal MRI , 2017, IEEE Transactions on Biomedical Engineering.

[18]  Zhiyuan Liu,et al.  Deep Learning Based Framework for Direct Reconstruction of PET Images , 2019, MICCAI.

[19]  Ning Guo,et al.  CT-guided PET parametric image reconstruction using deep neural network without prior training data , 2019, Medical Imaging.

[20]  Alexia Jolicoeur-Martineau,et al.  The relativistic discriminator: a key element missing from standard GAN , 2018, ICLR.

[21]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[22]  Junshen Xu,et al.  200x Low-dose PET Reconstruction using Deep Learning , 2017, ArXiv.

[23]  Yan Wang,et al.  Predicting standard-dose PET image from low-dose PET and multimodal MR images using mapping-based sparse representation , 2015, MLMI.