Iterative Reconstruction for Low-Dose CT using Deep Gradient Priors of Generative Model
IEEE Transactions on Radiation and Plasma Medical Sciences, Feb. 2022.
https://ieeexplore.ieee.org/abstract/document/9703672
The Code is created based on the method described in the following paper:
Iterative Reconstruction for Low-Dose CT using Deep Gradient Priors of Generative Model.
Zhuonan He, Yikun Zhang, Yu Guan, Bing Guan, Shanzhou Niu, Yi Zhang, Yang Chen, Qiegen Liu.
Dose reduction in computed tomography (CT) is essential for decreasing radiation risk in clinical applications. Iterative reconstruction is one of the most promising ways to compensate for the increased noise due to reduction of photon flux. Rather than most existing prior-driven algorithms that benefit from manually designed prior functions or supervised learning schemes, in this work we integrate the data-consistency as a conditional term into the iterative generative model for low-dose CT. At first, a score-based generative network is used for unsupervised distribution learning and the gradient of generative density prior is learned from normal-dose images. Then, the annealing Langevin dynamics is employed to update the trained priors with conditional scheme, i.e., the distance between the reconstructed image and the manifold is minimized along with data fidelity during reconstruction. Experimental comparisons demonstrated the noise reduction and detail preservation abilities of the proposed method.
Fig. 1. The training and reconstruction paradigm of the generative model-based algorithm EASEL. It consists of two components, i.e., a denoising score matching for score estimation involving various noise magnitudes simultaneously, and an iterative cycle for reconstruction including the annealed and conditional Langevin dynamics.
python==3.5
Pytorch==1.4.0
ODL==1.1.0
astra-toolbox==1.9.9dev
CUDA==9.0
python3.5 separate_ImageNet.py --model ncsn --runner Aapm_Runner_CTtest_10_noconv --config aapm_10C.yml --doc AapmCT_10C --test --image_folder output
The pretrained checkpoints can be download pretrained models from Baidu Drive. key number is "xt4l"
Fig. 2. Reconstruction results of AAPM challenge data for different methods. From left to right: reference image, FBP, TV, K-SVD, RED-CNN, DP-ResNet, EASEL.
-
Generative Modeling in Sinogram Domain for Sparse-view CT Reconstruction
[Paper] [Code] -
One Sample Diffusion Model in Projection Domain for Low-Dose CT Imaging
[Paper] [Code] -
REDAEP: Robust and Enhanced Denoising Autoencoding Prior for Sparse-View CT Reconstruction
[Paper] [Code] [PPT] -
Wavelet-improved score-based generative model for medical imaging
[Paper] -
Stage-by-stage Wavelet Optimization Refinement Diffusion Model for Sparse-view CT Reconstruction
[Paper] [Code] -
Dual-Domain Collaborative Diffusion Sampling for Multi-Source Stationary Computed Tomography Reconstruction
[Paper] [Code] -
Low-rank Angular Prior Guided Multi-diffusion Model for Few-shot Low-dose CT Reconstruction
[Paper] [Code] -
Physics-informed DeepCT: Sinogram Wavelet Decomposition Meets Masked Diffusion
[Paper] [Code] -
MSDiff: Multi-Scale Diffusion Model for Ultra-Sparse View CT Reconstruction
[Paper] [Code] -
Ordered-subsets Multi-diffusion Model for Sparse-view CT Reconstruction
[Paper] -
Virtual-mask Informed Prior for Sparse-view Dual-Energy CT Reconstruction
[Paper] [Code] -
Raw_data_generation [Code]
-
PRO: Projection Domain Synthesis for CT Imaging [Paper] [Code]
-
UniSino: Physics-Driven Foundational Model for Universal CT Sinogram Standardization[Paper] [Code]