Abstract:<h3> Purpose</h3><p>Deep‐neural‐network‐based image reconstruction has demonstrated promising performance in medical imaging for under‐sampled and low‐dose scenarios. However, it requires large amount of memory and extensive time for the training. It is especially challenging to train the reconstruction networks for three‐dimensional computed tomography (CT) because of the high resolution of CT images. The purpose of this work is to reduce the memory and time consumption of the training of the reconstruction networks for CT to make it practical for current hardware, while maintaining the quality of the reconstructed images.</p><h3> Methods</h3><p>We unrolled the proximal gradient descent algorithm for iterative image reconstruction to finite iterations and replaced the terms related to the penalty function with trainable convolutional neural networks (CNN). The network was trained greedily iteration by iteration in the image‐domain on patches, which requires reasonable amount of memory and time on mainstream graphics processing unit (GPU). To overcome the local‐minimum problem caused by greedy learning, we used deep UNet as the CNN and incorporated separable quadratic surrogate with ordered subsets for data fidelity, so that the solution could escape from easy local minimums and achieve better image quality.</p><h3> Results</h3><p>The proposed method achieved comparable image quality with state‐of‐the‐art neural network for CT image reconstruction on 2D sparse‐view and limited‐angle problems on the low‐dose CT challenge dataset. The difference in root‐mean‐square‐error (RMSE) and structural similarity index (SSIM) was within [−0.23,0.47]HU and [0,0.001] respectively with 95% confidence level. For 3D image reconstruction with ordinary‐size CT volume, the proposed method only needed 2 GB graphics processing unit (GPU) memory and 0.45 seconds per training iteration as minimum requirement, whereas existing methods may require 417 GB and 31 minutes. The proposed method achieved improved performance compared to total‐variation‐ and dictionary‐learning‐based iterative reconstruction for both 2D and 3D problems.</p><h3> Conclusions</h3><p>We proposed a training‐time computationally efficient neural network for CT image reconstruction. The proposed method achieved comparable image quality with state‐of‐the‐art neural network for CT reconstruction, with significantly reduced memory and time requirement during training. The proposed method is applicable to 3D image reconstruction problems such as cone‐beam CT and tomosynthesis on mainstream GPUs.</p><p>This article is protected by copyright. All rights reserved.</p>

Low-dose CT image reconstruction based on a deep neural network

Deep Fusion Network Based Sparse View CT Reconstructions for Clinical Diagnostic Scanners

Re-UNet: A Novel Multi-scale Reverse U-shaped Network Architecture for Low-dose CT Image Reconstruction

A Self-supervised Deep Learning Network for Low-Dose CT Reconstruction

Low Dose CT Image Reconstruction Using Deep Convolutional Residual Learning Network

Deep learning-based algorithms for low-dose CT imaging: A review

Low-dose CT denoising with convolutional neural network

Low-Dose CT Reconstruction Using Dataset-free Learning

Deep Learning-Based Low Dose Ct Imaging

Low-dose Dental CT Image Enhancement Using a Multiscale Feature Sensing Network

A deep convolutional neural network using directional wavelets for low-dose X-ray CT reconstruction

Enhancing Low-dose CT Image Reconstruction by Integrating Supervised and Unsupervised Learning

Investigation on Deep Learning Strategies for Reduced-dose CT Reconstruction

Slice-wise Reconstruction for Low-Dose Cone-Beam CT Using a Deep Residual Convolutional Neural Network

Low-Dose CT Image Enhancement Using Deep Learning

Noise-Generating and Imaging Mechanism Inspired Implicit Regularization Learning Network for Low Dose CT Reconstrution.

Re-UNet: a novel multi-scale reverse U-shape network architecture for low-dose CT image reconstruction

Computationally efficient deep neural network for computed tomography image reconstruction

200x Low-dose PET Reconstruction using Deep Learning

Can Deep Learning Outperform Modern Commercial CT Image Reconstruction Methods?

Combining Deep Learning and Adaptive Sparse Modeling for Low-dose CT Reconstruction