Iterative Low-dose CT Reconstruction with Priors Trained by Artificial Neural Network
A novel iterative reconstruction method combined with deep learning was proposed for low-dose CT imaging. Prior knowledge of the image was learned via k-sparse autoencoder (KSAE) from the normal-dose images. The KSAE was a multi-layer perceptron (MLP) which was trained by mapping normal-dose patches to themselves, whereas sparsity constraints on the hidden features were applied. The trained KSAE was then used as prior during iterative reconstruction, where the distance from the reconstructed image and the KSAE output was minimized along with the data fidelity term. Experiments demonstrated that the proposed method achieved better SSIM and CNR on quarter-dose CT images than dictionary learning and total variation. Once trained, the KSAE can be applied to reconstruction of different level of doses data.
Repository: https://github.com/wudufan/KSAERecon
Related publications:
https://ieeexplore.ieee.org/abstract/document/8038851
Low-dose CT Reconstruction with Noise2Noise Network and Testing-Time Fine-Tuning
A novel network-based prior was proposed for low-dose CT reconstruction. The projection data were split into the odd and even sets, where a network was used to map from the odd reconstruction to the even reconstruction and vice-versa. Due to the independent noises in the two projection sets, the network was able to remove noise but keep the structures. The network was built into an iterative reconstruction framework, where the reconstructed image was bounded by both the data fidelity and the network output. During the reconstruction, the network was optimized along with the reconstructed images. The network was trained by Noise2Noise, which did not require any normal-dose dataset to train. It achieved higher SSIM and better texture preservation compared to non-supervised methods such as non-local mean-constraint iterative reconstruction. Furthermore, the proposed iterative reconstruction also works with randomly initialized network, which enables fast denoising network prototyping for single set of data.
Repository: https://github.com/wudufan/Noise2NoiseReconstruction
Related publications:
https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/mp.15101
Low-dose CT reconstruction using spatially encoded nonlocal penalty
Purpose: Computed tomography (CT) is one of the most used imaging modalities for imaging both symptomatic and asymptomatic patients. However, because of the high demand for lower radiation dose during CT scans, the reconstructed image can suffer from noise and artifacts due to the trade-off between the image quality and the radiation dose. The purpose of this paper is to improve the image quality of quarter dose images and to select the best hyperparameters using the regular dose image as ground truth.
Methods: We first generated the axially stacked two-dimensional sinograms from the multislice raw projections with flying focal spots using a single slice rebinning method, which is an axially approximate method to provide simple implementation and efficient memory usage. To improve the image quality, a cost function containing the Poisson log-likelihood and spatially encoded nonlocal penalty is proposed. Specifically, an ordered subsets separable quadratic surrogates (OS-SQS) method for the log-likelihood is exploited and the patch-based similarity constraint with a spatially variant factor is developed to reduce the noise significantly while preserving features. Furthermore, we applied the Nesterov’s momentum method for acceleration and the diminishing number of subsets strategy for noise consistency. Fast nonlocal weight calculation is also utilized to reduce the computational cost.
Results: Datasets given by the Low Dose CT Grand Challenge were used for the validation, exploiting the training datasets with the regular and quarter dose data. The most important step in this paper was to fine-tune the hyperparameters to provide the best image for diagnosis. Using the regular dose filtered back-projection (FBP) image as ground truth, we could carefully select the hyperparameters by conducting a bias and standard deviation study, and we obtained the best images in a fixed number of iterations. We demonstrated that the proposed method with well selected hyperparameters improved the image quality using quarter dose data. The quarter dose proposed method was compared with the regular dose FBP, quarter dose FBP, and quarter dose l1-based 3-D TV method. We confirmed that the quarter dose proposed image was comparable to the regular dose FBP image and was better than images using other quarter dose methods. The reconstructed test images of the accreditation (ACR) CT phantom and 20 patients data were evaluated by radiologists at the Mayo-clinic, and this method was awarded first place in the Low Dose CT Grand Challenge.
Conclusion: We proposed the iterative CT reconstruction method using a spatially encoded nonlocal penalty and ordered subsets separable quadratic surrogates with the Nesterov’s momentum and diminishing number of subsets. The results demonstrated that the proposed method with fine-tuned hyperparameters can significantly improve the image quality and provide accurate diagnostic features at quarter dose.