Deep Convolutional Framelet Denosing for Low-Dose CT via Wavelet Residual Network

Potentials and caveats of AI in hybrid imaging

State-of-the-art patient management frequently mandates the investigation of both anatomy and physiology of the patients. Hybrid imaging modalities such as the PET/MRI, PET/CT and SPECT/CT have the ability to provide both structural and functional information of the investigated tissues in a single examination. With the introduction of such advanced hardware fusion, new problems arise such as the exceedingly large amount of multi-modality data that requires novel approaches of how to extract a maximum of clinical information from large sets of multi-dimensional imaging data. Artificial intelligence (AI) has emerged as one of the leading technologies that has shown promise in facilitating highly integrative analysis of multi-parametric data. Specifically, the usefulness of AI algorithms in the medical imaging field has been heavily investigated in the realms of (1) image acquisition and reconstruction, (2) post-processing and (3) data mining and modelling. Here, we aim to provide an overview of the challenges encountered in hybrid imaging and discuss how AI algorithms can facilitate potential solutions. In addition, we highlight the pitfalls and challenges in using advanced AI algorithms in the context of hybrid imaging and provide suggestions for building robust AI solutions that enable reproducible and transparent research.

Deep Learning for Ultrasound Localization Microscopy

By localizing microbubbles (MBs) in the vasculature, ultrasound localization microscopy (ULM) has recently been proposed, which greatly improves the spatial resolution of ultrasound (US) imaging and will be helpful for clinical diagnosis. Nevertheless, several challenges remain in fast ULM imaging. The main problems are that current localization methods used to implement fast ULM imaging, e.g., a previously reported localization method based on sparse recovery (CS-ULM), suffer from long data-processing time and exhaustive parameter tuning (optimization). To address these problems, in this paper, we propose a ULM method based on deep learning, which is achieved by using a modified sub-pixel convolutional neural network (CNN), termed as mSPCN-ULM. Simulations and in vivo experiments are performed to evaluate the performance of mSPCN-ULM. Simulation results show that even if under high-density condition (6.4 MBs/mm²), a high localization precision ( [Formula: see text] in the lateral direction and [Formula: see text] in the axial direction) and a high localization reliability (Jaccard index of 0.66) can be obtained by mSPCN-ULM, compared to CS-ULM. The in vivo experimental results indicate that with plane wave scan at a transmit center frequency of 15.625 MHz, microvessels with diameters of [Formula: see text] can be detected and adjacent microvessels with a distance of [Formula: see text] can be separated. Furthermore, when using GPU acceleration, the data-processing time of mSPCN-ULM can be shortened to ~6 sec/frame in the simulations and ~23 sec/frame in the in vivo experiments, which is 3-4 orders of magnitude faster than CS-ULM. Finally, once the network is trained, mSPCN-ULM does not need parameter tuning to implement ULM. As a result, mSPCN-ULM opens the door to implement ULM with fast data-processing speed, high imaging accuracy, short data-acquisition time, and high flexibility (robustness to parameters) characteristics.

Dense Recurrent Neural Networks for Accelerated MRI: History-Cognizant Unrolling of Optimization Algorithms

Inverse problems for accelerated MRI typically incorporate domain-specific knowledge about the forward encoding operator in a regularized reconstruction framework. Recently physics-driven deep learning (DL) methods have been proposed to use neural networks for data-driven regularization. These methods unroll iterative optimization algorithms to solve the inverse problem objective function, by alternating between domain-specific data consistency and data-driven regularization via neural networks. The whole unrolled network is then trained end-to-end to learn the parameters of the network. Due to simplicity of data consistency updates with gradient descent steps, proximal gradient descent (PGD) is a common approach to unroll physics-driven DL reconstruction methods. However, PGD methods have slow convergence rates, necessitating a higher number of unrolled iterations, leading to memory issues in training and slower reconstruction times in testing. Inspired by efficient variants of PGD methods that use a history of the previous iterates, we propose a history-cognizant unrolling of the optimization algorithm with dense connections across iterations for improved performance. In our approach, the gradient descent steps are calculated at a trainable combination of the outputs of all the previous regularization units. We also apply this idea to unrolling variable splitting methods with quadratic relaxation. Our results in reconstruction of the fastMRI knee dataset show that the proposed history-cognizant approach reduces residual aliasing artifacts compared to its conventional unrolled counterpart without requiring extra computational power or increasing reconstruction time.

Study of low-dose PET image recovery using supervised learning with CycleGAN

PET is a popular medical imaging modality for various clinical applications, including diagnosis and image-guided radiation therapy. The low-dose PET (LDPET) at a minimized radiation dosage is highly desirable in clinic since PET imaging involves ionizing radiation, and raises concerns about the risk of radiation exposure. However, the reduced dose of radioactive tracers could impact the image quality and clinical diagnosis. In this paper, a supervised deep learning approach with a generative adversarial network (GAN) and the cycle-consistency loss, Wasserstein distance loss, and an additional supervised learning loss, named as S-CycleGAN, is proposed to establish a non-linear end-to-end mapping model, and used to recover LDPET brain images. The proposed model, and two recently-published deep learning methods (RED-CNN and 3D-cGAN) were applied to 10% and 30% dose of 10 testing datasets, and a series of simulation datasets embedded lesions with different activities, sizes, and shapes. Besides vision comparisons, six measures including the NRMSE, SSIM, PSNR, LPIPS, SUVmax and SUVmean were evaluated for 10 testing datasets and 45 simulated datasets. Our S-CycleGAN approach had comparable SSIM and PSNR, slightly higher noise but a better perception score and preserving image details, much better SUVmean and SUVmax, as compared to RED-CNN and 3D-cGAN. Quantitative and qualitative evaluations indicate the proposed approach is accurate, efficient and robust as compared to other state-of-the-art deep learning methods.

Adaptive and Compressive Beamforming Using Deep Learning for Medical Ultrasound

In ultrasound (US) imaging, various types of adaptive beamforming techniques have been investigated to improve the resolution and the contrast-to-noise ratio of the delay and sum (DAS) beamformers. Unfortunately, the performance of these adaptive beamforming approaches degrades when the underlying model is not sufficiently accurate and the number of channels decreases. To address this problem, here, we propose a deep-learning-based beamformer to generate significantly improved images over widely varying measurement conditions and channel subsampling patterns. In particular, our deep neural network is designed to directly process full or subsampled radio frequency (RF) data acquired at various subsampling rates and detector configurations so that it can generate high-quality US images using a single beamformer. The origin of such input-dependent adaptivity is also theoretically analyzed. Experimental results using the B-mode focused US confirm the efficacy of the proposed methods.

Artificial intelligence in radiotherapy: a technological review

Radiation therapy (RT) is widely used to treat cancer. Technological advances in RT have occurred in the past 30 years. These advances, such as three-dimensional image guidance, intensity modulation, and robotics, created challenges and opportunities for the next breakthrough, in which artificial intelligence (AI) will possibly play important roles. AI will replace certain repetitive and labor-intensive tasks and improve the accuracy and consistency of others, particularly those with increased complexity because of technological advances. The improvement in efficiency and consistency is important to manage the increasing cancer patient burden to the society. Furthermore, AI may provide new functionalities that facilitate satisfactory RT. The functionalities include superior images for real-time intervention and adaptive and personalized RT. AI may effectively synthesize and analyze big data for such purposes. This review describes the RT workflow and identifies areas, including imaging, treatment planning, quality assurance, and outcome prediction, that benefit from AI. This review primarily focuses on deep-learning techniques, although conventional machine-learning techniques are also mentioned.

Incremental Image Noise Reduction in Coronary CT Angiography Using a Deep Learning-Based Technique with Iterative Reconstruction

Objective

To assess the feasibility of applying a deep learning-based denoising technique to coronary CT angiography (CCTA) along with iterative reconstruction for additional noise reduction.

Materials and Methods

We retrospectively enrolled 82 consecutive patients (male:female = 60:22; mean age, 67.0 ± 10.8 years) who had undergone both CCTA and invasive coronary artery angiography from March 2017 to June 2018. All included patients underwent CCTA with iterative reconstruction (ADMIRE level 3, Siemens Healthineers). We developed a deep learning based denoising technique (ClariCT.AI, ClariPI), which was based on a modified U-net type convolutional neural net model designed to predict the possible occurrence of low-dose noise in the originals. Denoised images were obtained by subtracting the predicted noise from the originals. Image noise, CT attenuation, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were objectively calculated. The edge rise distance (ERD) was measured as an indicator of image sharpness. Two blinded readers subjectively graded the image quality using a 5-point scale. Diagnostic performance of the CCTA was evaluated based on the presence or absence of significant stenosis (≥ 50% lumen reduction).

Results

Objective image qualities (original vs. denoised: image noise, 67.22 ± 25.74 vs. 52.64 ± 27.40; SNR [left main], 21.91 ± 6.38 vs. 30.35 ± 10.46; CNR [left main], 23.24 ± 6.52 vs. 31.93 ± 10.72; all p < 0.001) and subjective image quality (2.45 ± 0.62 vs. 3.65 ± 0.60, p < 0.001) improved significantly in the denoised images. The average ERDs of the denoised images were significantly smaller than those of originals (0.98 ± 0.08 vs. 0.09 ± 0.08, p < 0.001). With regard to diagnostic accuracy, no significant differences were observed among paired comparisons.

Conclusion

Application of the deep learning technique along with iterative reconstruction can enhance the noise reduction performance with a significant improvement in objective and subjective image qualities of CCTA images.

Microwave Imaging by Deep Learning Network: Feasibility and Training Method

Microwave image reconstruction based on a deep-learning method is investigated in this paper. The neural network is capable of converting measured microwave signals acquired from a 24×24 antenna array at 4 GHz into a 128×128 image. To reduce the training difficulty, we first developed an autoencoder by which high-resolution images (128×128) were represented with 256×1 vectors; then we developed the second neural network which aimed to map microwave signals to the compressed features (256×1 vector). Two neural networks can be combined to a full network to make reconstructions, when both are successfully developed. The present two-stage training method reduces the difficulty in training deep learning networks (DLN) for inverse reconstruction. The developed neural network is validated by simulation examples and experimental data with objects in different shapes/sizes, placed in different locations, and with dielectric constant ranging from 2~6. Comparisons between the imaging results achieved by the present method and two conventional approaches: distorted Born iterative method (DBIM) and phase confocal method (PCM) are also provided.

A Task-dependent Investigation on Dose and Texture in CT Image Reconstruction

Localizing and characterizing clinically-significant lung nodules, a potential precursor to lung cancer, at the lowest achievable radiation dose is demanded to minimize the stochastic radiation effects from x-ray computed tomography (CT). A minimal dose level is heavily dependent on the image reconstruction algorithms and clinical task, in which the tissue texture always plays an important role. This study aims to investigate the dependence through a task-based evaluation at multiple dose levels and variable textures in reconstructions with prospective patient studies. 133 patients with a suspicious pulmonary nodule scheduled for biopsy were recruited and the data was acquired at120kVp with three different dose levels of 100, 40 and 20mAs. Three reconstruction algorithms were implemented: analytical filtered back-projection (FBP) with optimal noise filtering; statistical Markov random field (MRF) model with optimal Huber weighting (MRF-H) for piecewise smooth reconstruction; and tissue-specific texture model (MRF-T) for texture preserved statistical reconstruction. Experienced thoracic radiologists reviewed and scored all images at random, blind to the CT dose and reconstruction algorithms. The radiologists identified the nodules in each image including the 133 biopsy target nodules and 66 other non-target nodules. For target nodule characterization, only MRF-T at 40mAs showed no statistically significant difference from FBP at 100mAs. For localizing both the target nodules and the non-target nodules, some as small as 3mm, MRF-T at 40 and 20mAs levels showed no statistically significant difference from FBP at 100mAs, respectively. MRF-H and FBP at 40 and 20mAs levels performed statistically differently from FBP at 100mAs. This investigation concluded that (1) the textures in the MRF-T reconstructions improves both the tasks of localizing and characterizing nodules at low dose CT and (2) the task of characterizing nodules is more challenging than the task of localizing nodules and needs more dose or enhanced textures from reconstruction.

Deep learning improves contrast in low-fluence photoacoustic imaging

Low fluence illumination sources can facilitate clinical transition of photoacoustic imaging because they are rugged, portable, affordable, and safe. However, these sources also decrease image quality due to their low fluence. Here, we propose a denoising method using a multi-level wavelet-convolutional neural network to map low fluence illumination source images to its corresponding high fluence excitation map. Quantitative and qualitative results show a significant potential to remove the background noise and preserve the structures of target. Substantial improvements up to 2.20, 2.25, and 4.3-fold for PSNR, SSIM, and CNR metrics were observed, respectively. We also observed enhanced contrast (up to 1.76-fold) in an in vivo application using our proposed methods. We suggest that this tool can improve the value of such sources in photoacoustic imaging.

DirectPET: full-size neural network PET reconstruction from sinogram data

Purpose: Neural network image reconstruction directly from measurement data is a relatively new field of research, which until now has been limited to producing small single-slice images (e.g., 1 × 128 × 128 ). We proposed a more efficient network design for positron emission tomography called DirectPET, which is capable of reconstructing multislice image volumes (i.e., 16 × 400 × 400 ) from sinograms. Approach: Large-scale direct neural network reconstruction is accomplished by addressing the associated memory space challenge through the introduction of a specially designed Radon inversion layer. Using patient data, we compare the proposed method to the benchmark ordered subsets expectation maximization (OSEM) algorithm using signal-to-noise ratio, bias, mean absolute error, and structural similarity measures. In addition, line profiles and full-width half-maximum measurements are provided for a sample of lesions. Results: DirectPET is shown capable of producing images that are quantitatively and qualitatively similar to the OSEM target images in a fraction of the time. We also report on an experiment where DirectPET is trained to map low-count raw data to normal count target images, demonstrating the method's ability to maintain image quality under a low-dose scenario. Conclusion: The ability of DirectPET to quickly reconstruct high-quality, multislice image volumes suggests potential clinical viability of the method. However, design parameters and performance boundaries need to be fully established before adoption can be considered.

One network to solve all ROIs: Deep learning CT for any ROI using differentiated backprojection

PURPOSE

Computed tomography for the reconstruction of region of interest (ROI) has advantages in reducing the x-ray dose and the use of a small detector. However, standard analytic reconstruction methods such as filtered back projection (FBP) suffer from severe cupping artifacts, and existing model-based iterative reconstruction methods require extensive computations. Recently, we proposed a deep neural network to learn the cupping artifacts, but the network was not generalized well for different ROIs due to the singularities in the corrupted images. Therefore, there is an increasing demand for a neural network that works well for any ROI size.

METHOD

Two types of neural networks are designed. The first type learns ROI size-specific cupping artifacts from FBP images, whereas the second type network is for the inversion of the truncated Hilbert transform from the truncated differentiated backprojection (DBP) data. Their generalizabilities for different ROI sizes, pixel sizes, detector pitch and starting angles for a short scan are then investigated.

RESULTS

Experimental results show that the new type of neural networks significantly outperform existing iterative methods for all ROI sizes despite significantly lower runtime complexity. In addition, performance improvement is consistent across different acquisition scenarios.

CONCLUSIONS

Since the proposed method consistently surpasses existing methods, it can be used as a general CT reconstruction engine for many practical applications without compromising possible detector truncation.

VVBP-Tensor in the FBP Algorithm: Its Properties and Application in Low-Dose CT Reconstruction

For decades, commercial X-ray computed tomography (CT) scanners have been using the filtered backprojection (FBP) algorithm for image reconstruction. However, the desire for lower radiation doses has pushed the FBP algorithm to its limit. Previous studies have made significant efforts to improve the results of FBP through preprocessing the sinogram, modifying the ramp filter, or postprocessing the reconstructed images. In this paper, we focus on analyzing and processing the stacked view-by-view backprojections (named VVBP-Tensor) in the FBP algorithm. A key challenge for our analysis lies in the radial structures in each backprojection slice. To overcome this difficulty, a sorting operation was introduced to the VVBP-Tensor in its z direction (the direction of the projection views). The results show that, after sorting, the tensor contains structures that are similar to those of the object, and structures in different slices of the tensor are correlated. We then analyzed the properties of the VVBP-Tensor, including structural self-similarity, tensor sparsity, and noise statistics. Considering these properties, we have developed an algorithm using the tensor singular value decomposition (named VVBP-tSVD) to denoise the VVBP-Tensor for low-mAs CT imaging. Experiments were conducted using a physical phantom and clinical patient data with different mAs levels. The results demonstrate that the VVBP-tSVD is superior to all competing methods under different reconstruction schemes, including sinogram preprocessing, image postprocessing, and iterative reconstruction. We conclude that the VVBP-Tensor is a suitable processing target for improving the quality of FBP reconstruction, and the proposed VVBP-tSVD is an effective algorithm for noise reduction in low-mAs CT imaging. This preliminary work might provide a heuristic perspective for reviewing and rethinking the FBP algorithm.

SPULTRA: Low-Dose CT Image Reconstruction With Joint Statistical and Learned Image Models

Low-dose CT image reconstruction has been a popular research topic in recent years. A typical reconstruction method based on post-log measurements is called penalized weighted-least squares (PWLS). Due to the underlying limitations of the post-log statistical model, the PWLS reconstruction quality is often degraded in low-dose scans. This paper investigates a shifted-Poisson (SP) model based likelihood function that uses the pre-log raw measurements that better represents the measurement statistics, together with a data-driven regularizer exploiting a Union of Learned TRAnsforms (SPULTRA). Both the SP induced data-fidelity term and the regularizer in the proposed framework are nonconvex. The proposed SPULTRA algorithm uses quadratic surrogate functions for the SP induced data-fidelity term. Each iteration involves a quadratic subproblem for updating the image, and a sparse coding and clustering subproblem that has a closed-form solution. The SPULTRA algorithm has a similar computational cost per iteration as its recent counterpart PWLS-ULTRA that uses post-log measurements, and it provides better image reconstruction quality than PWLS-ULTRA, especially in low-dose scans.

TomoGAN: low-dose synchrotron x-ray tomography with generative adversarial networks: discussion

Synchrotron-based x-ray tomography is a noninvasive imaging technique that allows for reconstructing the internal structure of materials at high spatial resolutions from tens of micrometers to a few nanometers. In order to resolve sample features at smaller length scales, however, a higher radiation dose is required. Therefore, the limitation on the achievable resolution is set primarily by noise at these length scales. We present TomoGAN, a denoising technique based on generative adversarial networks, for improving the quality of reconstructed images for low-dose imaging conditions. We evaluate our approach in two photon-budget-limited experimental conditions: (1) sufficient number of low-dose projections (based on Nyquist sampling), and (2) insufficient or limited number of high-dose projections. In both cases, the angular sampling is assumed to be isotropic, and the photon budget throughout the experiment is fixed based on the maximum allowable radiation dose on the sample. Evaluation with both simulated and experimental datasets shows that our approach can significantly reduce noise in reconstructed images, improving the structural similarity score of simulation and experimental data from 0.18 to 0.9 and from 0.18 to 0.41, respectively. Furthermore, the quality of the reconstructed images with filtered back projection followed by our denoising approach exceeds that of reconstructions with the simultaneous iterative reconstruction technique, showing the computational superiority of our approach.

ADAPTIVE-NET: deep computed tomography reconstruction network with analytical domain transformation knowledge

Background

Recently, the paradigm of computed tomography (CT) reconstruction has shifted as the deep learning technique evolves. In this study, we proposed a new convolutional neural network (called ADAPTIVE-NET) to perform CT image reconstruction directly from a sinogram by integrating the analytical domain transformation knowledge.

Methods

In the proposed ADAPTIVE-NET, a specific network layer with constant weights was customized to transform the sinogram onto the CT image domain via analytical back-projection. With this new framework, feature extractions were performed simultaneously on both the sinogram domain and the CT image domain. The Mayo low dose CT (LDCT) data was used to validate the new network. In particular, the new network was compared with the previously proposed residual encoder-decoder (RED)-CNN network. For each network, the mean square error (MSE) loss with and without VGG-based perceptual loss was compared. Furthermore, to evaluate the image quality with certain metrics, the noise correlation was quantified via the noise power spectrum (NPS) on the reconstructed LDCT for each method.

Results

CT images that have clinically relevant dimensions of 512×512 can be easily reconstructed from a sinogram on a single graphics processing unit (GPU) with moderate memory size (e.g., 11 GB) by ADAPTIVE-NET. With the same MSE loss function, the new network is able to generate better results than the RED-CNN. Moreover, the new network is able to reconstruct natural looking CT images with enhanced image quality if jointly using the VGG loss.

Conclusions

The newly proposed end-to-end supervised ADAPTIVE-NET is able to reconstruct high-quality LDCT images directly from a sinogram.

Image Quality and Lesion Detection on Deep Learning Reconstruction and Iterative Reconstruction of Submillisievert Chest and Abdominal CT

OBJECTIVE. The objective of this study was to compare image quality and clinically significant lesion detection on deep learning reconstruction (DLR) and iterative reconstruction (IR) images of submillisievert chest and abdominopelvic CT. MATERIALS AND METHODS. Our prospective multiinstitutional study included 59 adult patients (33 women, 26 men; mean age ± SD, 65 ± 12 years old; mean body mass index [weight in kilograms divided by the square of height in meters] = 27 ± 5) who underwent routine chest (n = 22; 16 women, six men) and abdominopelvic (n = 37; 17 women, 20 men) CT on a 640-MDCT scanner (Aquilion ONE, Canon Medical Systems). All patients gave written informed consent for the acquisition of low-dose (LD) CT (LDCT) after a clinically indicated standard-dose (SD) CT (SDCT). The SDCT series (120 kVp, 164-644 mA) were reconstructed with interactive reconstruction (IR) (adaptive iterative dose reduction [AIDR] 3D, Canon Medical Systems), and the LDCT (100 kVp, 120 kVp; 30-50 mA) were reconstructed with filtered back-projection (FBP), IR (AIDR 3D and forward-projected model-based iterative reconstruction solution [FIRST], Canon Medical Systems), and deep learning reconstruction (DLR) (Advanced Intelligent Clear-IQ Engine [AiCE], Canon Medical Systems). Four subspecialty-trained radiologists first read all LD image sets and then compared them side-by-side with SD AIDR 3D images in an independent, randomized, and blinded fashion. Subspecialty radiologists assessed image quality of LDCT images on a 3-point scale (1 = unacceptable, 2 = suboptimal, 3 = optimal). Descriptive statistics were obtained, and the Wilcoxon sign rank test was performed. RESULTS. Mean volume CT dose index and dose-length product for LDCT (2.1 ± 0.8 mGy, 49 ± 13mGy·cm) were lower than those for SDCT (13 ± 4.4 mGy, 567 ± 249 mGy·cm) (p < 0.0001). All 31 clinically significant abdominal lesions were seen on SD AIDR 3D and LD DLR images. Twenty-five, 18, and seven lesions were detected on LD AIDR 3D, LD FIRST, and LD FBP images, respectively. All 39 pulmonary nodules detected on SD AIDR 3D images were also noted on LD DLR images. LD DLR images were deemed acceptable for interpretation in 97% (35/37) of abdominal and 95-100% (21-22/22) of chest LDCT studies (p = 0.2-0.99). The LD FIRST, LD AIDR 3D, and LD FBP images had inferior image quality compared with SD AIDR 3D images (p < 0.0001). CONCLUSION. At submillisievert chest and abdominopelvic CT doses, DLR enables image quality and lesion detection superior to commercial IR and FBP images.

Image Reconstruction: From Sparsity to Data-adaptive Methods and Machine Learning

The field of medical image reconstruction has seen roughly four types of methods. The first type tended to be analytical methods, such as filtered back-projection (FBP) for X-ray computed tomography (CT) and the inverse Fourier transform for magnetic resonance imaging (MRI), based on simple mathematical models for the imaging systems. These methods are typically fast, but have suboptimal properties such as poor resolution-noise trade-off for CT. A second type is iterative reconstruction methods based on more complete models for the imaging system physics and, where appropriate, models for the sensor statistics. These iterative methods improved image quality by reducing noise and artifacts. The FDA-approved methods among these have been based on relatively simple regularization models. A third type of methods has been designed to accommodate modified data acquisition methods, such as reduced sampling in MRI and CT to reduce scan time or radiation dose. These methods typically involve mathematical image models involving assumptions such as sparsity or low-rank. A fourth type of methods replaces mathematically designed models of signals and systems with data-driven or adaptive models inspired by the field of machine learning. This paper focuses on the two most recent trends in medical image reconstruction: methods based on sparsity or low-rank models, and data-driven methods based on machine learning techniques.

CNN-based PET sinogram repair to mitigate defective block detectors

Positron emission tomography (PET) scanners continue to increase sensitivity and axial coverage by adding an ever expanding array of block detectors. As they age, one or more block detectors may lose sensitivity due to a malfunction or component failure. The sinogram data missing as a result thereof can lead to artifacts and other image degradations. We propose to mitigate the effects of malfunctioning block detectors by carrying out sinogram repair using a deep convolutional neural network. Experiments using whole-body patient studies with varying amounts of raw data removed are used to show that the neural network significantly outperforms previously published methods with respect to normalized mean squared error for raw sinograms, a multi-scale structural similarity measure for reconstructed images and with regard to quantitative accuracy.

Domain Progressive 3D Residual Convolution Network to Improve Low-Dose CT Imaging

The wide applications of X-ray computed tomography (CT) bring low-dose CT (LDCT) into a clinical prerequisite, but reducing the radiation exposure in CT often leads to significantly increased noise and artifacts, which might lower the judgment accuracy of radiologists. In this paper, we put forward a domain progressive 3D residual convolution network (DP-ResNet) for the LDCT imaging procedure that contains three stages: sinogram domain network (SD-net), filtered back projection (FBP), and image domain network (ID-net). Though both are based on the residual network structure, the SD-net and ID-net provide complementary effect on improving the final LDCT quality. The experimental results with both simulated and real projection data show that this domain progressive deep-learning network achieves significantly improved performance by combing the network processing in the two domains.

Physics-informed Deep Learning for Dual-Energy Computed Tomography Image Processing

Dual-energy CT (DECT) was introduced to address the inability of conventional single-energy computed tomography (SECT) to distinguish materials with similar absorbances but different elemental compositions. However, material decomposition algorithms based purely on the physics of the underlying attenuation process have several limitations, leading to low signal-to-noise ratio (SNR) in the derived material-specific images. To overcome these, we trained a convolutional neural network (CNN) to develop a framework to reconstruct non-contrast SECT images from DECT scans. We show that the traditional physics-based decomposition algorithms do not bring to bear the full information content of the image data. A CNN that leverages the underlying physics of the DECT image generation process as well as the anatomic information gleaned via training with actual images can generate higher fidelity processed DECT images.

Learning to Reconstruct Computed Tomography Images Directly From Sinogram Data Under A Variety of Data Acquisition Conditions

Computed tomography (CT) is widely used in medical diagnosis and non-destructive detection. Image reconstruction in CT aims to accurately recover pixel values from measured line integrals, i.e., the summed pixel values along straight lines. Provided that the acquired data satisfy the data sufficiency condition as well as other conditions regarding the view angle sampling interval and the severity of transverse data truncation, researchers have discovered many solutions to accurately reconstruct the image. However, if these conditions are violated, accurate image reconstruction from line integrals remains an intellectual challenge. In this paper, a deep learning method with a common network architecture, termed iCT-Net, was developed and trained to accurately reconstruct images for previously solved and unsolved CT reconstruction problems with high quantitative accuracy. Particularly, accurate reconstructions were achieved for the case when the sparse view reconstruction problem (i.e., compressed sensing problem) is entangled with the classical interior tomographic problems.

A performance comparison of convolutional neural network-based image denoising methods: The effect of loss functions on low-dose CT images

PURPOSE

Convolutional neural network (CNN)-based image denoising techniques have shown promising results in low-dose CT denoising. However, CNN often introduces blurring in denoised images when trained with a widely used pixel-level loss function. Perceptual loss and adversarial loss have been proposed recently to further improve the image denoising performance. In this paper, we investigate the effect of different loss functions on image denoising performance using task-based image quality assessment methods for various signals and dose levels.

METHODS

We used a modified version of U-net that was effective at reducing the correlated noise in CT images. The loss functions used for comparison were two pixel-level losses (i.e., the mean-squared error and the mean absolute error), Visual Geometry Group network-based perceptual loss (VGG loss), adversarial loss used to train the Wasserstein generative adversarial network with gradient penalty (WGAN-GP), and their weighted summation. Each image denoising method was applied to reconstructed images and sinogram images independently and validated using the extended cardiac-torso (XCAT) simulation and Mayo Clinic datasets. In the XCAT simulation, we generated fan-beam CT datasets with four different dose levels (25%, 50%, 75%, and 100% of a normal-dose level) using 10 XCAT phantoms and inserted signals in a test set. The signals had two different shapes (spherical and spiculated), sizes (4 and 12 mm), and contrast levels (60 and 160 HU). To evaluate signal detectability, we used a detection task SNR (tSNR) calculated from a non-prewhitening model observer with an eye filter. We also measured the noise power spectrum (NPS) and modulation transfer function (MTF) to compare the noise and signal transfer properties.

RESULTS

Compared to CNNs without VGG loss, VGG-loss-based CNNs achieved a more similar tSNR to that of the normal-dose CT for all signals at different dose levels except for a small signal at the 25% dose level. For a low-contrast signal at 25% or 50% dose, adding other losses to the VGG loss showed more improved performance than only using VGG loss. The NPS shapes from VGG-loss-based CNN closely matched that of normal-dose CT images while CNN without VGG loss overly reduced the mid-high-frequency noise power at all dose levels. MTF also showed VGG-loss-based CNN with better-preserved high resolution for all dose and contrast levels. It is also observed that additional WGAN-GP loss helps improve the noise and signal transfer properties of VGG-loss-based CNN.

CONCLUSIONS

The evaluation results using tSNR, NPS, and MTF indicate that VGG-loss-based CNNs are more effective than those without VGG loss for natural denoising of low-dose images and WGAN-GP loss improves the denoising performance of VGG-loss-based CNNs, which corresponds with the qualitative evaluation.

Competitive performance of a modularized deep neural network compared to commercial algorithms for low-dose CT image reconstruction

Commercial iterative reconstruction techniques help to reduce CT radiation dose but altered image appearance and artifacts limit their adoptability and potential use. Deep learning has been investigated for low-dose CT (LDCT). Here we design a modularized neural network for LDCT and compared it with commercial iterative reconstruction methods from three leading CT vendors. While popular networks are trained for an end-to-end mapping, our network performs an end-to-process mapping so that intermediate denoised images are obtained with associated noise reduction directions towards a final denoised image. The learned workflow allows radiologists-in-the-loop to optimize the denoising depth in a task-specific fashion. Our network was trained with the Mayo LDCT Dataset, and tested on separate chest and abdominal CT exams from Massachusetts General Hospital. The best deep learning reconstructions were systematically compared to the best iterative reconstructions in a double-blinded reader study. This study confirms that our deep learning approach performed either favorably or comparably in terms of noise suppression and structural fidelity, and is much faster than the commercial iterative reconstruction algorithms.

Cycle-consistent adversarial denoising network for multiphase coronary CT angiography

PURPOSE

In multiphase coronary CT angiography (CTA), a series of CT images are taken at different levels of radiation dose during the examination. Although this reduces the total radiation dose, the image quality during the low-dose phases is significantly degraded. Recently, deep neural network approaches based on supervised learning technique have demonstrated impressive performance improvement over conventional model-based iterative methods for low-dose CT. However, matched low- and routine-dose CT image pairs are difficult to obtain in multiphase CT. To address this problem, we aim at developing a new deep learning framework.

METHOD

We propose an unsupervised learning technique that can remove the noise of the CT images in the low-dose phases by learning from the CT images in the routine dose phases. Although a supervised learning approach is not applicable due to the differences in the underlying heart structure in two phases, the images are closely related in two phases, so we propose a cycle-consistent adversarial denoising network to learn the mapping between the low- and high-dose cardiac phases.

RESULTS

Experimental results showed that the proposed method effectively reduces the noise in the low-dose CT image while preserving detailed texture and edge information. Moreover, thanks to the cyclic consistency and identity loss, the proposed network does not create any artificial features that are not present in the input images. Visual grading and quality evaluation also confirm that the proposed method provides significant improvement in diagnostic quality.

CONCLUSIONS

The proposed network can learn the image distributions from the routine-dose cardiac phases, which is a big advantage over the existing supervised learning networks that need exactly matched low- and routine-dose CT images. Considering the effectiveness and practicability of the proposed method, we believe that the proposed can be applied for many other CT acquisition protocols.

Framing U-Net via Deep Convolutional Framelets: Application to Sparse-View CT

X-ray computed tomography (CT) using sparse projection views is a recent approach to reduce the radiation dose. However, due to the insufficient projection views, an analytic reconstruction approach using the filtered back projection (FBP) produces severe streaking artifacts. Recently, deep learning approaches using large receptive field neural networks such as U-Net have demonstrated impressive performance for sparse-view CT reconstruction. However, theoretical justification is still lacking. Inspired by the recent theory of deep convolutional framelets, the main goal of this paper is, therefore, to reveal the limitation of U-Net and propose new multi-resolution deep learning schemes. In particular, we show that the alternative U-Net variants such as dual frame and tight frame U-Nets satisfy the so-called frame condition which makes them better for effective recovery of high frequency edges in sparse-view CT. Using extensive experiments with real patient data set, we demonstrate that the new network architectures provide better reconstruction performance.