Intelligent Parameter Tuning in Optimization-Based Iterative CT Reconstruction via Deep Reinforcement Learning

Learning-Assisted Fast Determination of Regularization Parameter in Constrained Image Reconstruction

OBJECTIVE

To leverage machine learning (ML) for fast selection of optimal regularization parameter in constrained image reconstruction.

METHODS

Constrained image reconstruction is often formulated as a regularization problem and selecting a good regularization parameter value is an essential step. We solved this problem using an ML-based approach by leveraging the finding that for a specific constrained reconstruction problem defined for a fixed class of image functions, the optimal regularization parameter value is weakly subject-dependent and the dependence can be captured using few experimental data. The proposed method has four key steps: a) solution of a given constrained reconstruction problem for a few (say, 3) pre-selected regularization parameter values, b) extraction of multiple approximated quality metrics from the initial reconstructions, c) predicting the true quality metrics values from the approximated values using pre-trained neural networks, and d) determination of the optimal regularization parameter by fusing the predicted quality metrics.

RESULTS

The effectiveness of the proposed method was demonstrated in two constrained reconstruction problems. Compared with L-curve-based method, the proposed method determined the regularization parameters much faster and produced substantially improved reconstructions. Our method also outperformed state-of-the-art learning-based methods when trained with limited experimental data.

CONCLUSION

This paper demonstrates the feasibility and improved reconstruction quality by using machine learning to determine the regularization parameter in constrained reconstruction.

SIGNIFICANCE

The proposed method substantially reduces the computational burden of the traditional methods (e.g., L-curve) or relaxes the requirement of large training data by modern learning-based methods, thus enhancing the practical utility of constrained reconstruction.

Adaptive scatter kernel deconvolution modeling for cone-beam CT scatter correction via deep reinforcement learning

BACKGROUND

Scattering photons can seriously contaminate cone-beam CT (CBCT) image quality with severe artifacts and substantial degradation of CT value accuracy, which is a major concern limiting the widespread application of CBCT in the medical field. The scatter kernel deconvolution (SKD) method commonly used in clinic requires a Monte Carlo (MC) simulation to determine numerous quality-related kernel parameters, and it cannot realize intelligent scatter kernel parameter optimization, causing limited accuracy of scatter estimation.

PURPOSE

Aiming at improving the scatter estimation accuracy of the SKD algorithm, an intelligent scatter correction framework integrating the SKD with deep reinforcement learning (DRL) scheme is proposed.

METHODS

Our method firstly builds a scatter kernel model to iteratively convolve with raw projections, and then the deep Q-network of the DRL scheme is introduced to intelligently interact with the scatter kernel to achieve a projection adaptive parameter optimization. The potential of the proposed framework is demonstrated on CBCT head and pelvis simulation data and experimental CBCT measurement data. Furthermore, we have implemented the U-net based scatter estimation approach for comparison.

RESULTS

The simulation study demonstrates that the mean absolute percentage error (MAPE) of the proposed method is less than 9.72% and the peak signal-to-noise ratio (PSNR) is higher than 23.90 dB, while for the conventional SKD algorithm, the minimum MAPE is 17.92% and the maximum PSNR is 19.32 dB. In the measurement study, we adopt a hardware-based beam stop array algorithm to obtain the scatter-free projections as a comparison baseline, and our method can achieve superior performance with MAPE < 17.79% and PSNR > 16.34 dB.

CONCLUSIONS

In this paper, we propose an intelligent scatter correction framework that integrates the physical scatter kernel model with DRL algorithm, which has the potential to improve the accuracy of the clinical scatter correction method to obtain better CBCT imaging quality.

Adaptive prior image constrained total generalized variation for low-dose dynamic cerebral perfusion CT reconstruction

BACKGROUND

Dynamic cerebral perfusion CT (DCPCT) can provide valuable insight into cerebral hemodynamics by visualizing changes in blood within the brain. However, the associated high radiation dose of the standard DCPCT scanning protocol has been a great concern for the patient and radiation physics. Minimizing the x-ray exposure to patients has been a major effort in the DCPCT examination. A simple and cost-effective approach to achieve low-dose DCPCT imaging is to lower the x-ray tube current in data acquisition. However, the image quality of low-dose DCPCT will be degraded because of the excessive quantum noise.

OBJECTIVE

To obtain high-quality DCPCT images, we present a statistical iterative reconstruction (SIR) algorithm based on penalized weighted least squares (PWLS) using adaptive prior image constrained total generalized variation (APICTGV) regularization (PWLS-APICTGV).

METHODS

APICTGV regularization uses the precontrast scanned high-quality CT image as an adaptive structural prior for low-dose PWLS reconstruction. Thus, the image quality of low-dose DCPCT is improved while essential features of targe image are well preserved. An alternating optimization algorithm is developed to solve the cost function of the PWLS-APICTGV reconstruction.

RESULTS

PWLS-APICTGV algorithm was evaluated using a digital brain perfusion phantom and patient data. Compared to other competing algorithms, the PWLS-APICTGV algorithm shows better noise reduction and structural details preservation. Furthermore, the PWLS-APICTGV algorithm can generate more accurate cerebral blood flow (CBF) map than that of other reconstruction methods.

CONCLUSIONS

PWLS-APICTGV algorithm can significantly suppress noise while preserving the important features of the reconstructed DCPCT image, thus achieving a great improvement in low-dose DCPCT imaging.

Reducing the risk of hallucinations with interpretable deep learning models for low-dose CT denoising: comparative performance analysis

Objective.Reducing CT radiation dose is an often proposed measure to enhance patient safety, which, however results in increased image noise, translating into degradation of clinical image quality. Several deep learning methods have been proposed for low-dose CT (LDCT) denoising. The high risks posed by possible hallucinations in clinical images necessitate methods which aid the interpretation of deep learning networks. In this study, we aim to use qualitative reader studies and quantitative radiomics studies to assess the perceived quality, signal preservation and statistical feature preservation of LDCT volumes denoised by deep learning. We aim to compare interpretable deep learning methods with classical deep neural networks in clinical denoising performance.Approach.We conducted an image quality analysis study to assess the image quality of the denoised volumes based on four criteria to assess the perceived image quality. We subsequently conduct a lesion detection/segmentation study to assess the impact of denoising on signal detectability. Finally, a radiomic analysis study was performed to observe the quantitative and statistical similarity of the denoised images to standard dose CT (SDCT) images.Main results.The use of specific deep learning based algorithms generate denoised volumes which are qualitatively inferior to SDCT volumes(p< 0.05). Contrary to previous literature, denoising the volumes did not reduce the accuracy of the segmentation (p> 0.05). The denoised volumes, in most cases, generated radiomics features which were statistically similar to those generated from SDCT volumes (p> 0.05).Significance.Our results show that the denoised volumes have a lower perceived quality than SDCT volumes. Noise and denoising do not significantly affect detectability of the abdominal lesions. Denoised volumes also contain statistically identical features to SDCT volumes.

Similarity-Driven Fine-Tuning Methods for Regularization Parameter Optimization in PET Image Reconstruction

We present an adaptive method for fine-tuning hyperparameters in edge-preserving regularization for PET image reconstruction. For edge-preserving regularization, in addition to the smoothing parameter that balances data fidelity and regularization, one or more control parameters are typically incorporated to adjust the sensitivity of edge preservation by modifying the shape of the penalty function. Although there have been efforts to develop automated methods for tuning the hyperparameters in regularized PET reconstruction, the majority of these methods primarily focus on the smoothing parameter. However, it is challenging to obtain high-quality images without appropriately selecting the control parameters that adjust the edge preservation sensitivity. In this work, we propose a method to precisely tune the hyperparameters, which are initially set with a fixed value for the entire image, either manually or using an automated approach. Our core strategy involves adaptively adjusting the control parameter at each pixel, taking into account the degree of patch similarities calculated from the previous iteration within the pixel's neighborhood that is being updated. This approach allows our new method to integrate with a wide range of existing parameter-tuning techniques for edge-preserving regularization. Experimental results demonstrate that our proposed method effectively enhances the overall reconstruction accuracy across multiple image quality metrics, including peak signal-to-noise ratio, structural similarity, visual information fidelity, mean absolute error, root-mean-square error, and mean percentage error.

Deep image and feature prior algorithm based on U-ConformerNet structure

PURPOSE

The reconstruction performance of the deep image prior (DIP) approach is limited by the conventional convolutional layer structure and it is difficult to enhance its potential. In order to improve the quality of image reconstruction and suppress artifacts, we propose a DIP algorithm with better performance, and verify its superiority in the latest case.

METHODS

We construct a new U-ConformerNet structure as the DIP algorithm's network, replacing the traditional convolutional layer-based U-net structure, and introduce the 'lpips' deep network based feature distance regularization method. Our algorithm can switch between supervised and unsupervised modes at will to meet different needs.

RESULTS

The reconstruction was performed on the low dose CT dataset (LoDoPaB). Our algorithm attained a PSNR of more than 35 dB under unsupervised conditions, and the PSNR under the supervised condition is greater than 36 dB. Both of which are better than the performance of the DIP-TV. Furthermore, the accuracy of this method is positively connected with the quality of the a priori image with the help of deep networks. In terms of noise eradication and artifact suppression, the DIP algorithm with U-ConformerNet structure outperforms the standard DIP method based on convolutional structure.

CONCLUSIONS

It is known by experimental verification that, in unsupervised mode, the algorithm improves the output PSNR by at least 2-3 dB when compared to the DIP-TV algorithm (proposed in 2020). In supervised mode, our algorithm approaches that of the state-of-the-art end-to-end deep learning algorithms.

Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey

Reinforcement learning takes sequential decision-making approaches by learning the policy through trial and error based on interaction with the environment. Combining deep learning and reinforcement learning can empower the agent to learn the interactions and the distribution of rewards from state-action pairs to achieve effective and efficient solutions in more complex and dynamic environments. Deep reinforcement learning (DRL) has demonstrated astonishing performance in surpassing the human-level performance in the game domain and many other simulated environments. This paper introduces the basics of reinforcement learning and reviews various categories of DRL algorithms and DRL models developed for medical image analysis and radiation treatment planning optimization. We will also discuss the current challenges of DRL and approaches proposed to make DRL more generalizable and robust in a real-world environment. DRL algorithms, by fostering the designs of the reward function, agents interactions and environment models, can resolve the challenges from scarce and heterogeneous annotated medical image data, which has been a major obstacle to implementing deep learning models in the clinic. DRL is an active research area with enormous potential to improve deep learning applications in medical imaging and radiation therapy planning.

Deep-Learning-Based Ultrasound Sound-Speed Tomography Reconstruction with Tikhonov Pseudo-Inverse Priori

Ultrasound sound-speed tomography (USST) is a promising technology for breast imaging and breast cancer detection. Its reconstruction is a complex non-linear mapping from the projection data to the sound-speed image (SSI). The traditional reconstruction methods include mainly the ray-based methods and the waveform-based methods. The ray-based methods with linear approximation have low computational cost but low reconstruction quality; the full wave-based methods with the complex non-linear model have high quality but high cost. To achieve both high quality and low cost, we introduced traditional linear approximation as prior knowledge into a deep neural network and treated the complex non-linear mapping of USST reconstruction as a combination of linear mapping and non-linear mapping. In the proposed method, the linear mapping was seamlessly implemented with a fully connected layer and initialized using the Tikhonov pseudo-inverse matrix. The non-linear mapping was implemented using a U-shape Net (U-Net). Furthermore, we proposed the Tikhonov U-shape net (TU-Net), in which the linear mapping was done before the non-linear mapping, and the U-shape Tikhonov net (UT-Net), in which the non-linear mapping was done before the linear mapping. Moreover, we conducted simulations and experiments for evaluation. In the numerical simulation, the root-mean-squared error was 6.49 and 4.29 m/s for the UT-Net and TU-Net, the peak signal-to-noise ratio was 49.01 and 52.90 dB, the structural similarity was 0.9436 and 0.9761 and the reconstruction time was 10.8 and 11.3 ms, respectively. In this study, the SSIs obtained with the proposed methods exhibited high sound-speed accuracy. Both the UT-Net and the TU-Net achieved high quality and low computational cost.

Basis and current state of computed tomography perfusion imaging: a review

Computed tomography perfusion (CTP) is a functional imaging that allows for providing capillary-level hemodynamics information of the desired tissue in clinics. In this paper, we aim to offer insight into CTP imaging which covers the basics and current state of CTP imaging, then summarize the technical applications in the CTP imaging as well as the future technological potential. At first, we focus on the fundamentals of CTP imaging including systematically summarized CTP image acquisition and hemodynamic parameter map estimation techniques. A short assessment is presented to outline the clinical applications with CTP imaging, and then a review of radiation dose effect of the CTP imaging on the different applications is presented. We present a categorized methodology review on known and potential solvable challenges of radiation dose reduction in CTP imaging. To evaluate the quality of CTP images, we list various standardized performance metrics. Moreover, we present a review on the determination of infarct and penumbra. Finally, we reveal the popularity and future trend of CTP imaging.

The use of deep learning methods in low-dose computed tomography image reconstruction: a systematic review

Conventional reconstruction techniques, such as filtered back projection (FBP) and iterative reconstruction (IR), which have been utilised widely in the image reconstruction process of computed tomography (CT) are not suitable in the case of low-dose CT applications, because of the unsatisfying quality of the reconstructed image and inefficient reconstruction time. Therefore, as the demand for CT radiation dose reduction continues to increase, the use of artificial intelligence (AI) in image reconstruction has become a trend that attracts more and more attention. This systematic review examined various deep learning methods to determine their characteristics, availability, intended use and expected outputs concerning low-dose CT image reconstruction. Utilising the methodology of Kitchenham and Charter, we performed a systematic search of the literature from 2016 to 2021 in Springer, Science Direct, arXiv, PubMed, ACM, IEEE, and Scopus. This review showed that algorithms using deep learning technology are superior to traditional IR methods in noise suppression, artifact reduction and structure preservation, in terms of improving the image quality of low-dose reconstructed images. In conclusion, we provided an overview of the use of deep learning approaches in low-dose CT image reconstruction together with their benefits, limitations, and opportunities for improvement.

Limited parameter denoising for low-dose X-ray computed tomography using deep reinforcement learning

BACKGROUND

The use of deep learning has successfully solved several problems in the field of medical imaging. Deep learning has been applied to the CT denoising problem successfully. However, the use of deep learning requires large amounts of data to train deep convolutional networks (CNNs). Moreover, due to large parameter count, such deep CNNs may cause unexpected results.

PURPOSE

In this study, we introduce a novel CT denoising framework, which has interpretable behaviour, and provides useful results with limited data.

METHODS

We employ bilateral filtering in both the projection and volume domains to remove noise. To account for non-stationary noise, we tune the σ parameters of the volume for every projection view, and for every volume pixel. The tuning is carried out by two deep CNNs. Due to impracticality of labelling, the two deep CNNs are trained via a Deep-Q reinforcement learning task. The reward for the task is generated by using a custom reward function represented by a neural network. Our experiments were carried out on abdominal scans for the Mayo Clinic TCIA dataset, and the AAPM Low Dose CT Grand Challenge.

RESULTS

Our denoising framework has excellent denoising performance increasing the PSNR from 28.53 to 28.93, and increasing the SSIM from 0.8952 to 0.9204. We outperform several state-of-the-art deep CNNs, which have several orders of magnitude higher number of parameters (p-value (PSNR) = 0.000, p-value (SSIM) = 0.000). Our method does not introduce any blurring, which is introduced by MSE loss based methods, or any deep learning artifacts, which are introduced by WGAN based models. Our ablation studies show that parameter tuning and using our reward network results in the best possible results.

CONCLUSIONS

We present a novel CT denoising framework, which focuses on interpretability to deliver good denoising performance, especially with limited data. Our method outperforms state-of-the-art deep neural networks. Future work will be focused on accelerating our method, and generalizing to different geometries and body parts. This article is protected by copyright. All rights reserved.

Convex optimization algorithms in medical image reconstruction-in the age of AI

The past decade has seen the rapid growth of model based image reconstruction (MBIR) algorithms, which are often applications or adaptations of convex optimization algorithms from the optimization community. We review some state-of-the-art algorithms that have enjoyed wide popularity in medical image reconstruction, emphasize known connections between different algorithms, and discuss practical issues such as computation and memory cost. More recently, deep learning (DL) has forayed into medical imaging, where the latest development tries to exploit the synergy between DL and MBIR to elevate the MBIR's performance. We present existing approaches and emerging trends in DL-enhanced MBIR methods, with particular attention to the underlying role of convexity and convex algorithms on network architecture. We also discuss how convexity can be employed to improve the generalizability and representation power of DL networks in general.

Artificial Intelligence in Radiation Therapy

Artificial intelligence (AI) has great potential to transform the clinical workflow of radiotherapy. Since the introduction of deep neural networks, many AI-based methods have been proposed to address challenges in different aspects of radiotherapy. Commercial vendors have started to release AI-based tools that can be readily integrated to the established clinical workflow. To show the recent progress in AI-aided radiotherapy, we have reviewed AI-based studies in five major aspects of radiotherapy including image reconstruction, image registration, image segmentation, image synthesis, and automatic treatment planning. In each section, we summarized and categorized the recently published methods, followed by a discussion of the challenges, concerns, and future development. Given the rapid development of AI-aided radiotherapy, the efficiency and effectiveness of radiotherapy in the future could be substantially improved through intelligent automation of various aspects of radiotherapy.

Hierarchical Granular Computing-Based Model and Its Reinforcement Structural Learning for Construction of Long-Term Prediction Intervals

As one of the most essential sources of energy, byproduct gas plays a pivotal role in the steel industry, for which the flow tendency is generally regarded as the guidance for planning and scheduling in real production. In order to obtain the numeric estimation along with its reliability, the construction of prediction intervals (PIs) is highly demanded by any practical applications as well as being long term for providing more information on future trends. Bearing this in mind, in this article, a hierarchical granular computing (HGrC)-based model is established for constructing long-term PIs, in which probabilistic modeling gives rise to a long horizon of numeric prediction, and the deployment of information granularities hierarchically extends the result to be interval-valued format. Considering that the structure of this model has a direct impact on its performance, Monte-Carlo search with a policy gradient technique is then applied for reinforcement structure learning. Compared with the existing methods, the size (length) of the granules in the proposed approach is unequal so that it becomes effective for not only periodic but also nonperiodic data. Furthermore, with the use of parallel strategy, the efficiency can be also guaranteed for real-world applications. The experimental results demonstrate that the proposed method is superior to other commonly encountered techniques, and the stability of the structure learning process behaves better when compared with other reinforcement learning approaches.

Review and Prospect: Artificial Intelligence in Advanced Medical Imaging

Artificial intelligence (AI) as an emerging technology is gaining momentum in medical imaging. Recently, deep learning-based AI techniques have been actively investigated in medical imaging, and its potential applications range from data acquisition and image reconstruction to image analysis and understanding. In this review, we focus on the use of deep learning in image reconstruction for advanced medical imaging modalities including magnetic resonance imaging (MRI), computed tomography (CT), and positron emission tomography (PET). Particularly, recent deep learning-based methods for image reconstruction will be emphasized, in accordance with their methodology designs and performances in handling volumetric imaging data. It is expected that this review can help relevant researchers understand how to adapt AI for medical imaging and which advantages can be achieved with the assistance of AI.

Patient-specific hyperparameter learning for optimization-based CT image reconstruction

We propose a hyperparameter learning framework that learnspatient-specifichyperparameters for optimization-based image reconstruction problems for x-ray CT applications. The framework consists of two functional modules: (1) a hyperparameter learning module parameterized by a convolutional neural network, (2) an image reconstruction module that takes as inputs both the noisy sinogram and the hyperparameters from (1) and generates the reconstructed images. As a proof-of-concept study, in this work we focus on a subclass of optimization-based image reconstruction problems with exactly computable solutions so that the whole network can be trained end-to-end in an efficient manner. Unlike existing hyperparameter learning methods, our proposed framework generates patient-specific hyperparameters from the sinogram of the same patient. Numerical studies demonstrate the effectiveness of our proposed approach compared to bi-level optimization.

A prior image constraint robust principal component analysis reconstruction method for sparse segmental multi-energy computed tomography

Background

Multi-energy computed tomography (MECT) is a promising technique in medical imaging, especially for quantitative imaging. However, high technical requirements and system costs barrier its step into clinical practice.

Methods

We propose a novel sparse segmental MECT (SSMECT) scheme and corresponding reconstruction method, which is a cost-efficient way to realize MECT on a conventional single-source CT. For the data acquisition, the X-ray source is controlled to maintain an energy within a segmental arc, and then switch alternately to another energy level. This scan only needs to switch tube voltage a few times to acquire multi-energy data, but leads to sparse-view and limited-angle issues in image reconstruction. To solve this problem, we propose a prior image constraint robust principal component analysis (PIC-RPCA) reconstruction method, which introduces structural similarity and spectral correlation into the reconstruction.

Results

A numerical simulation and a real phantom experiment were conducted to demonstrate the efficacy and robustness of the scan scheme and reconstruction method. The results showed that our proposed reconstruction method could have achieved better multi-energy images than other competing methods both quantitatively and qualitatively.

Conclusions

Our proposed SSMECT scan with PIC-RPCA reconstruction method could lower kVp switching frequency while achieving satisfactory reconstruction accuracy and image quality.

Image quality guided iterative reconstruction for low-dose CT based on CT image statistics

Iterative reconstruction framework shows predominance in low dose and incomplete data situation. In the iterative reconstruction framework, there are two components, i.e., fidelity term aims to maintain the structure details of the reconstructed object, and the regularization term uses prior information to suppress the artifacts such as noise. A regularization parameter balances them, aiming to find a good trade-off between noise and resolution. Currently, the regularization parameters are selected as a rule of thumb or some prior knowledge assumption is required, which limits practical uses. Furthermore, the computation cost of regularization parameter selection is also heavy. In this paper, we address this problem by introducing CT image quality assessment (IQA) into the iterative reconstruction framework. Several steps are involved during the study. First, we analyze the CT image statistics using the dual dictionary method. Regularities are observed and concluded, revealing the relationship among the regularization parameter, iterations, and CT image quality. Second, with derivation and simplification of DDL procedure, a CT IQA metric named SODVAC is designed. The SODVAC locates the optimal regularization parameter that results in the reconstructed image with distinct structures and with no noise or little noise. Third, we introduce SODVAC into the iterative reconstruction framework and then propose a general image-quality-guided iterative reconstruction (QIR) framework and give a specific framework example (sQIR) by introducing SODVAC into the iterative reconstruction framework. sQIR simultaneously optimizes the reconstructed image and the regularization parameter during the iterations. Results confirm the effectiveness of the proposed method. No prior information needed and low computation cost are the advantages of our method compared with existing state-of-theart L-curve and ZIP selection strategies.

Deep reinforcement learning in medical imaging: A literature review

Deep reinforcement learning (DRL) augments the reinforcement learning framework, which learns a sequence of actions that maximizes the expected reward, with the representative power of deep neural networks. Recent works have demonstrated the great potential of DRL in medicine and healthcare. This paper presents a literature review of DRL in medical imaging. We start with a comprehensive tutorial of DRL, including the latest model-free and model-based algorithms. We then cover existing DRL applications for medical imaging, which are roughly divided into three main categories: (i) parametric medical image analysis tasks including landmark detection, object/lesion detection, registration, and view plane localization; (ii) solving optimization tasks including hyperparameter tuning, selecting augmentation strategies, and neural architecture search; and (iii) miscellaneous applications including surgical gesture segmentation, personalized mobile health intervention, and computational model personalization. The paper concludes with discussions of future perspectives.

Noise2Context: Context-assisted learning 3D thin-layer for low-dose CT

PURPOSE

Computed tomography (CT) has played a vital role in medical diagnosis, assessment, and therapy planning, etc. In clinical practice, concerns about the increase of x-ray radiation exposure attract more and more attention. To lower the x-ray radiation, low-dose CT (LDCT) has been widely adopted in certain scenarios, while it will induce the degradation of CT image quality. In this paper, we proposed a deep learning-based method that can train denoising neural networks without any clean data.

METHODS

In this work, for 3D thin-slice LDCT scanning, we first drive an unsupervised loss function which was equivalent to a supervised loss function with paired noisy and clean samples when the noise in the different slices from a single scan was uncorrelated and zero-mean. Then, we trained the denoising neural network to map one noise LDCT image to its two adjacent LDCT images in a single 3D thin-layer LDCT scanning, simultaneously. In essence, with some latent assumptions, we proposed an unsupervised loss function to train the denoising neural network in an unsupervised manner, which integrated the similarity between adjacent CT slices in 3D thin-layer LDCT.

RESULTS

Further experiments on Mayo LDCT dataset and a realistic pig head were carried out. In the experiments using Mayo LDCT dataset, our unsupervised method can obtain performance comparable to that of the supervised baseline. With the realistic pig head, our method can achieve optimal performance at different noise levels as compared to all the other methods that demonstrated the superiority and robustness of the proposed Noise2Context.

CONCLUSIONS

In this work, we present a generalizable LDCT image denoising method without any clean data. As a result, our method not only gets rid of the complex artificial image priors but also amounts of paired high-quality training datasets.

Autonomous Collision Avoidance Using MPC with LQR-Based Weight Transformation

Model predictive control (MPC) is a multi-objective control technique that can handle system constraints. However, the performance of an MPC controller highly relies on a proper prioritization weight for each objective, which highlights the need for a precise weight tuning technique. In this paper, we propose an analytical tuning technique by matching the MPC controller performance with the performance of a linear quadratic regulator (LQR) controller. The proposed methodology derives the transformation of a LQR weighting matrix with a fixed weighting factor using a discrete algebraic Riccati equation (DARE) and designs an MPC controller using the idea of a discrete time linear quadratic tracking problem (LQT) in the presence of constraints. The proposed methodology ensures optimal performance between unconstrained MPC and LQR controllers and provides a sub-optimal solution while the constraints are active during transient operations. The resulting MPC behaves as the discrete time LQR by selecting an appropriate weighting matrix in the MPC control problem and ensures the asymptotic stability of the system. In this paper, the effectiveness of the proposed technique is investigated in the application of a novel vehicle collision avoidance system that is designed in the form of linear inequality constraints within MPC. The simulation results confirm the potency of the proposed MPC control technique in performing a safe, feasible and collision-free path while respecting the inputs, states and collision avoidance constraints.

Iterative reconstruction for low-dose cerebral perfusion computed tomography using prior image induced diffusion tensor

Cerebral perfusion computed tomography (CPCT) can depict the functional status of cerebral circulation at the tissue level; hence, it has been increasingly used to diagnose patients with cerebrovascular disease. However, there is a significant concern that CPCT scanning protocol could expose patients to excessive radiation doses. Although reducing the x-ray tube current when acquiring CPCT projection data is an effective method for reducing radiation dose, this technique usually results in degraded image quality. To enhance the image quality of low-dose CPCT, we present a prior image induced diffusion tensor (PIDT) for statistical iterative reconstruction, based on the penalized weighted least-squares (PWLS) criterion, which we referred to as PWLS-PIDT, for simplicity. Specifically, PIDT utilizes the geometric features of pre-contrast scanned high-quality CT image as a structure prior for PWLS reconstruction; therefore, the low-dose CPCT images are enhanced while preserving important features in the target image. An effective alternating minimization algorithm is developed to solve the associated objective function in the PWLS-PIDT reconstruction. We conduct qualitative and quantitative studies to evaluate the PWLS-PIDT reconstruction with a digital brain perfusion phantom and patient data. With this method, the noise in the reconstructed CPCT images is more substantially reduced than that of other competing methods, without sacrificing structural details significantly. Furthermore, the CPCT sequential images reconstructed via the PWLS-PIDT method can derive more accurate hemodynamic parameter maps than those of other competing methods.

SPECTnet: a deep learning neural network for SPECT image reconstruction

Background

Single photon emission computed tomography (SPECT) is an important functional tool for clinical diagnosis and scientific research of brain disorders, but suffers from limited spatial resolution and high noise due to hardware design and imaging physics. The present study is to develop a deep learning technique for SPECT image reconstruction that directly converts raw projection data to image with high resolution and low noise, while an efficient training method specifically applicable to medical image reconstruction is presented.

Methods

A custom software was developed to generate 20,000 2-D brain phantoms, of which 16,000 were used to train the neural network, 2,000 for validation, and the final 2,000 for testing. To reduce development difficulty, a two-step training strategy for network design was adopted. We first compressed full-size activity image (128×128 pixels) to a one-D vector consisting of 256×1 pixels, accomplished by an autoencoder (AE) consisting of an encoder and a decoder. The vector is a good representation of the full-size image in a lower-dimensional space and was used as a compact label to develop the second network that maps between the projection-data domain and the vector domain. Since the label had 256 pixels only, the second network was compact and easy to converge. The second network, when successfully developed, was connected to the decoder (a portion of AE) to decompress the vector to a regular 128×128 image. Therefore, a complex network was essentially divided into two compact neural networks trained separately in sequence but eventually connectable.

Results

A total of 2,000 test examples, a synthetic brain phantom, and de-identified patient data were used to validate SPECTnet. Results obtained from SPECTnet were compared with those obtained from our clinic OS-EM method. Images with lower noise and more accurate information in the uptake areas were obtained by SPECTnet.

Conclusions

The challenge of developing a complex deep neural network is reduced by training two separate compact connectable networks. The combination of the two networks forms the full version of SPECTnet. Results show that the developed neural network can produce more accurate SPECT images.

Simultaneous needle catheter selection and dwell time optimization for preplanning of high-dose-rate brachytherapy of prostate cancer

PURPOSE

Needle catheter positions critically affect the quality of treatment plans in prostate cancer high-dose-rate (HDR) brachytherapy. The current standard needle positioning approach is based on human intuition, which cannot guarantee a high-quality plan. This study proposed a method to simultaneously select needle catheter positions and determine dwell time for preplanning of HDR brachytherapy of prostate cancer.

METHODS

We formulated the needle catheter selection problem and inverse dwell time optimization problem in a unified framework. In addition to the dose objectives of the planning target volume (PTV) and organs at risk (OARs), the objective function incorporated a group-sparsity term with a needle-specific adaptive weighting scheme to generate high-quality plans with the minimal number of needle catheters. The optimization problem was solved by a fast-iterative shrinkage-thresholding algorithm. For validation purposes, we tested the proposed algorithm on 10 patient cases previously treated at our institution and compared the resulting plans with plans generated using needle catheters selected manually.

RESULTS

Compared to the plan with manually selected needle catheters, when normalizing both plans to the same PTV coverage V _100% = 95%, the plans generated by the proposed algorithm reduced median V _125% from 65% to 64%, but increased median V _150% from 35% to 38%, and V _200% from 14% to 16%. All planning objectives were met. All clinically important dosimetric parameters of OARs were reduced. D _1cc of bladder and rectum were reduced from 8.57 Gy to 8.50 Gy and from 7.24 Gy to 6.80 Gy, respectively. D _max of urethra was reduced from 15.85 Gy to 15.77 Gy. The median number of selected needle catheters was reduced by two. The computational time for solving the proposed optimization problem was ∼90 s using MATLAB.

CONCLUSION

The proposed algorithm was able to generate plans for prostate cancer HDR brachytherapy preplanning with increased median conformity index (0.73-0.77) and slightly lower median homogeneity index (0.64-0.62) with the number of selected needles reduced by two compared to the manual needle selection approach.

Artificial intelligence and machine learning for medical imaging: A technology review

Artificial intelligence (AI) has recently become a very popular buzzword, as a consequence of disruptive technical advances and impressive experimental results, notably in the field of image analysis and processing. In medicine, specialties where images are central, like radiology, pathology or oncology, have seized the opportunity and considerable efforts in research and development have been deployed to transfer the potential of AI to clinical applications. With AI becoming a more mainstream tool for typical medical imaging analysis tasks, such as diagnosis, segmentation, or classification, the key for a safe and efficient use of clinical AI applications relies, in part, on informed practitioners. The aim of this review is to present the basic technological pillars of AI, together with the state-of-the-art machine learning methods and their application to medical imaging. In addition, we discuss the new trends and future research directions. This will help the reader to understand how AI methods are now becoming an ubiquitous tool in any medical image analysis workflow and pave the way for the clinical implementation of AI-based solutions.

Improving efficiency of training a virtual treatment planner network via knowledge-guided deep reinforcement learning for intelligent automatic treatment planning of radiotherapy

PURPOSE

We previously proposed an intelligent automatic treatment planning framework for radiotherapy, in which a virtual treatment planner network (VTPN) is built using deep reinforcement learning (DRL) to operate a treatment planning system (TPS) by adjusting treatment planning parameters in it to generate high-quality plans. We demonstrated the potential feasibility of this idea in prostate cancer intensity-modulated radiation therapy (IMRT). Despite the success, the process to train a VTPN via the standard DRL approach with an ϵ-greedy algorithm was time-consuming. The required training time was expected to grow with the complexity of the treatment planning problem, preventing the development of VTPN for more complicated but clinically relevant scenarios. In this study, we proposed a novel knowledge-guided DRL (KgDRL) approach that incorporated knowledge from human planners to guide the training process to improve the efficiency of training a VTPN.

METHOD

Using prostate cancer IMRT as a test bed, we first summarized a number of rules in the actions of adjusting treatment planning parameters of our in-house TPS. During the training process of VTPN, in addition to randomly navigating the large state-action space, as in the standard DRL approach using the ϵ-greedy algorithm, we also sampled actions defined by the rules. The priority of sampling actions from rules decreased over the training process to encourage VTPN to explore new policy on parameter adjustment that were not covered by the rules. To test this idea, we trained a VTPN using KgDRL and compared its performance with another VTPN trained using the standard DRL approach. Both networks were trained using 10 training patient cases and five additional cases for validation, while another 59 cases were employed for the evaluation purpose.

RESULTS

It was found that both VTPNs trained via KgDRL and standard DRL spontaneously learned how to operate the in-house TPS to generate high-quality plans, achieving plan quality scores of 8.82 (±0.29) and 8.43 (±0.48), respectively. Both VTPNs outperformed treatment planning purely based on the rules, which had a plan score of 7.81 (±1.59). VTPN trained with eight episodes using KgDRL was able to perform similar to that trained using DRL with 100 epochs. The training time was reduced from more than a week to ~13 hrs.

CONCLUSION

The proposed KgDRL framework was effective in accelerating the training process of a VTPN by incorporating human knowledge, which will facilitate the development of VTPN for more complicated treatment planning scenarios.

Ant Colony-Based Hyperparameter Optimisation in Total Variation Reconstruction in X-ray Computed Tomography

In this paper, a computer-aided training method for hyperparameter selection of limited data X-ray computed tomography (XCT) reconstruction was proposed. The proposed method employed the ant colony optimisation (ACO) approach to assist in hyperparameter selection for the adaptive-weighted projection-controlled steepest descent (AwPCSD) algorithm, which is a total-variation (TV) based regularisation algorithm. During the implementation, there was a colony of artificial ants that swarm through the AwPCSD algorithm. Each ant chose a set of hyperparameters required for its iterative CT reconstruction and the correlation coefficient (CC) score was given for reconstructed images compared to the reference image. A colony of ants in one generation left a pheromone through its chosen path representing a choice of hyperparameters. Higher score means stronger pheromones/probabilities to attract more ants in the next generations. At the end of the implementation, the hyperparameter configuration with the highest score was chosen as an optimal set of hyperparameters. In the experimental results section, the reconstruction using hyperparameters from the proposed method was compared with results from three other cases: the conjugate gradient least square (CGLS), the AwPCSD algorithm using the set of arbitrary hyperparameters and the cross-validation method.The experiments showed that the results from the proposed method were superior to those of the CGLS algorithm and the AwPCSD algorithm using the set of arbitrary hyperparameters. Although the results of the ACO algorithm were slightly inferior to those of the cross-validation method as measured by the quantitative metrics, the ACO algorithm was over 10 times faster than cross—Validation. The optimal set of hyperparameters from the proposed method was also robust against an increase of noise in the data and can be applicable to different imaging samples with similar context. The ACO approach in the proposed method was able to identify optimal values of hyperparameters for a dataset and, as a result, produced a good quality reconstructed image from limited number of projection data. The proposed method in this work successfully solves a problem of hyperparameters selection, which is a major challenge in an implementation of TV based reconstruction algorithms.

Reviewing Machine Learning and Image Processing Based Decision-Making Systems for Breast Cancer Imaging

Breast cancer (BC) is the leading cause of death among women worldwide. It affects in general women older than 40 years old. Medical images analysis is one of the most promising research areas since it provides facilities for diagnosis and decision-making of several diseases such as BC. This paper conducts a Structured Literature Review (SLR) of the use of Machine Learning (ML) and Image Processing (IP) techniques to deal with BC imaging. A set of 530 papers published between 2000 and August 2019 were selected and analyzed according to ten criteria: year and publication channel, empirical type, research type, medical task, machine learning techniques, datasets used, validation methods, performance measures and image processing techniques which include image pre-processing, segmentation, feature extraction and feature selection. Results showed that diagnosis was the most used medical task and that Deep Learning techniques (DL) were largely used to perform classification. Furthermore, we found out that classification was the most ML objective investigated followed by prediction and clustering. Most of the selected studies used Mammograms as imaging modalities rather than Ultrasound or Magnetic Resonance Imaging with the use of public or private datasets with MIAS as the most frequently investigated public dataset. As for image processing techniques, the majority of the selected studies pre-process their input images by reducing the noise and normalizing the colors, and some of them use segmentation to extract the region of interest with the thresholding method. For feature extraction, we note that researchers extracted the relevant features using classical feature extraction techniques (e.g. Texture features, Shape features, etc.) or DL techniques (e. g. VGG16, VGG19, ResNet, etc.), and finally few papers used feature selection techniques in particular the filter methods.

Full-Spectrum-Knowledge-Aware Tensor Model for Energy-Resolved CT Iterative Reconstruction

Energy-resolved computed tomography (ErCT) with a photon counting detector concurrently produces multiple CT images corresponding to different photon energy ranges. It has the potential to generate energy-dependent images with improved contrast-to-noise ratio and sufficient material-specific information. Since the number of detected photons in one energy bin in ErCT is smaller than that in conventional energy-integrating CT (EiCT), ErCT images are inherently more noisy than EiCT images, which leads to increased noise and bias in the subsequent material estimation. In this work, we first deeply analyze the intrinsic tensor properties of two-dimensional (2D) ErCT images acquired in different energy bins and then present a F ull- S pectrum-knowledge-aware Tensor analysis and processing (FSTensor) method for ErCT reconstruction to suppress noise-induced artifacts to obtain high-quality ErCT images and high-accuracy material images. The presented method is based on three considerations: (1) 2D ErCT images obtained in different energy bins can be treated as a 3-order tensor with three modes, i.e., width, height and energy bin, and a rich global correlation exists among the three modes, which can be characterized by tensor decomposition. (2) There is a locally piecewise smooth property in the 3-order ErCT images, and it can be captured by a tensor total variation regularization. (3) The images from the full spectrum are much better than the ErCT images with respect to noise variance and structural details and serve as external information to improve the reconstruction performance. We then develop an alternating direction method of multipliers algorithm to numerically solve the presented FSTensor method. We further utilize a genetic algorithm to tackle the parameter selection in ErCT reconstruction, instead of manually determining parameters. Simulation, preclinical and synthesized clinical ErCT results demonstrate that the presented FSTensor method leads to significant improvements over the filtered back-projection, robust principal component analysis, tensor-based dictionary learning and low-rank tensor decomposition with spatial-temporal total variation methods.

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.

Machine learning techniques for biomedical image segmentation: An overview of technical aspects and introduction to state-of-art applications

In recent years, significant progress has been made in developing more accurate and efficient machine learning algorithms for segmentation of medical and natural images. In this review article, we highlight the imperative role of machine learning algorithms in enabling efficient and accurate segmentation in the field of medical imaging. We specifically focus on several key studies pertaining to the application of machine learning methods to biomedical image segmentation. We review classical machine learning algorithms such as Markov random fields, k-means clustering, random forest, etc. Although such classical learning models are often less accurate compared to the deep-learning techniques, they are often more sample efficient and have a less complex structure. We also review different deep-learning architectures, such as the artificial neural networks (ANNs), the convolutional neural networks (CNNs), and the recurrent neural networks (RNNs), and present the segmentation results attained by those learning models that were published in the past 3 yr. We highlight the successes and limitations of each machine learning paradigm. In addition, we discuss several challenges related to the training of different machine learning models, and we present some heuristics to address those challenges.

Operating a treatment planning system using a deep-reinforcement learning-based virtual treatment planner for prostate cancer intensity-modulated radiation therapy treatment planning

PURPOSE

In the treatment planning process of intensity-modulated radiation therapy (IMRT), a human planner operates the treatment planning system (TPS) to adjust treatment planning parameters, for example, dose volume histogram (DVH) constraints' locations and weights, to achieve a satisfactory plan for each patient. This process is usually time-consuming, and the plan quality depends on planer's experience and available planning time. In this study, we proposed to model the behaviors of human planners in treatment planning by a deep reinforcement learning (DRL)-based virtual treatment planner network (VTPN), such that it can operate the TPS in a human-like manner for treatment planning.

METHODS AND MATERIALS

Using prostate cancer IMRT as an example, we established the VTPN using a deep neural network developed. We considered an in-house optimization engine with a weighted quadratic objective function. Virtual treatment planner network was designed to observe an intermediate plan DVHs and decide the action to improve the plan by changing weights and threshold dose in the objective function. We trained the VTPN in an end-to-end DRL process in 10 patient cases. A plan score was used to measure plan quality. We demonstrated the feasibility and effectiveness of the trained VTPN in another 64 patient cases.

RESULTS

Virtual treatment planner network was trained to spontaneously learn how to adjust treatment planning parameters to generate high-quality treatment plans. In the 64 testing cases, with initialized parameters, quality score was 4.97 (±2.02), with 9.0 being the highest possible score. Using VTPN to perform treatment planning improved quality score to 8.44 (±0.48).

CONCLUSIONS

To our knowledge, this was the first time that intelligent treatment planning behaviors of human planner in external beam IMRT are autonomously encoded in an artificial intelligence system. The trained VTPN is capable of behaving in a human-like way to produce high-quality plans.

An adaptive cooperation with reinforcement learning for robot soccer games

A strategy system with self-improvement and self-learning abilities for robot soccer system has been developed in this study. This work focuses on the cooperation strategy for the task assignment and develops an adaptive cooperation method for this system. This method was inspired by reinforcement learning (RL) and game theory. The developed system includes two subsystems: the task assignment system and the RL system. The task assignment system assigns one of the four roles, Attacker, Helper, Defender, and Goalkeeper, to each separate robot with the same physical and mechanical conditions to achieve cooperation. The assigned role to robots considers the situation in the game field. Each role has its own behaviors and tasks. The RL helps the Helper and Defender to improve the ability of their policy selection on the real-time confrontation. The RL system can not only learn to figure up how Helper helps its teammates to form an attack or a defense type but also learn to stand a proper defensive strategy. Some experiments on FIRE simulator and standard platform have been demonstrated that the proposed method performs better than the competitors.

An introduction to deep learning in medical physics: advantages, potential, and challenges

As one of the most popular approaches in artificial intelligence, deep learning (DL) has attracted a lot of attention in the medical physics field over the past few years. The goals of this topical review article are twofold. First, we will provide an overview of the method to medical physics researchers interested in DL to help them start the endeavor. Second, we will give in-depth discussions on the DL technology to make researchers aware of its potential challenges and possible solutions. As such, we divide the article into two major parts. The first part introduces general concepts and principles of DL and summarizes major research resources, such as computational tools and databases. The second part discusses challenges faced by DL, present available methods to mitigate some of these challenges, as well as our recommendations.

Statistical image-based material decomposition for triple-energy computed tomography using total variation regularization

BACKGROUND

Triple-energy computed tomography (TECT) can obtain x-ray attenuation measurements at three energy spectra, thereby allowing identification of different material compositions with same or very similar attenuation coefficients. This ability is known as material decomposition, which can decompose TECT images into different basis material image. However, the basis material image would be severely degraded when material decomposition is directly performed on the noisy TECT measurements using a matrix inversion method.

OBJECTIVE

To achieve high quality basis material image, we present a statistical image-based material decomposition method for TECT, which uses the penalized weighted least-squares (PWLS) criteria with total variation (TV) regularization (PWLS-TV).

METHODS

The weighted least-squares term involves the noise statistical properties of the material decomposition process, and the TV regularization penalizes differences between local neighboring pixels in a decomposed image, thereby contributing to improving the quality of the basis material image. Subsequently, an alternating optimization method is used to minimize the objective function.

RESULTS

The performance of PWLS-TV is quantitatively evaluated using digital and mouse thorax phantoms. The experimental results show that PWLS-TV material decomposition method can greatly improve the quality of decomposed basis material image compared to the quality of images obtained using the competing methods in terms of suppressing noise and preserving edge and fine structure details.

CONCLUSIONS

The PWLS-TV method can simultaneously perform noise reduction and material decomposition in one iterative step, and it results in a considerable improvement of basis material image quality.

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.

Residual convolutional neural network for predicting response of transarterial chemoembolization in hepatocellular carcinoma from CT imaging

Background

We attempted to train and validate a model of deep learning for the preoperative prediction of the response of patients with intermediate-stage hepatocellular carcinoma (HCC) undergoing transarterial chemoembolization (TACE).

Method

All computed tomography (CT) images were acquired for 562 patients from the Nan Fang Hospital (NFH), 89 patients from Zhu Hai Hospital Affiliated with Jinan University (ZHHAJU), and 138 patients from the Sun Yat-sen University Cancer Center (SYUCC). We built a predictive model from the outputs using the transfer learning techniques of a residual convolutional neural network (ResNet50). The prediction accuracy for each patch was revaluated in two independent validation cohorts.

Results

In the training set (NFH), the deep learning model had an accuracy of 84.3% and areas under curves (AUCs) of 0.97, 0.96, 0.95, and 0.96 for complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD), respectively. In the other two validation sets (ZHHAJU and SYUCC), the deep learning model had accuracies of 85.1% and 82.8% for CR, PR, SD, and PD. The ResNet50 model also had high AUCs for predicting the objective response of TACE therapy in patches and patients of three cohorts. Decision curve analysis (DCA) showed that the ResNet50 model had a high net benefit in the two validation cohorts.

Conclusion

The deep learning model presented a good performance for predicting the response of TACE therapy and could help clinicians in better screening patients with HCC who can benefit from the interventional treatment.

Key Points

• Therapy response of TACE can be predicted by a deep learning model based on CT images.

• The probability value from a trained or validation deep learning model showed significant correlation with different therapy responses.

• Further improvement is necessary before clinical utilization.

Electronic supplementary material

The online version of this article (10.1007/s00330-019-06318-1) contains supplementary material, which is available to authorized users.

Intelligent inverse treatment planning via deep reinforcement learning, a proof-of-principle study in high dose-rate brachytherapy for cervical cancer

Inverse treatment planning in radiation therapy is formulated as solving optimization problems. The objective function and constraints consist of multiple terms designed for different clinical and practical considerations. Weighting factors of these terms are needed to define the optimization problem. While a treatment planning optimization engine can solve the optimization problem with given weights, adjusting the weights to yield a high-quality plan is typically performed by a human planner. Yet the weight-tuning task is labor intensive, time consuming, and it critically affects the final plan quality. An automatic weight-tuning approach is strongly desired. The procedure of weight adjustment to improve the plan quality is essentially a decision-making problem. Motivated by the tremendous success in deep learning for decision making with human-level intelligence, we propose a novel framework to adjust the weights in a human-like manner. This study used inverse treatment planning in high-dose-rate brachytherapy (HDRBT) for cervical cancer as an example. We developed a weight-tuning policy network (WTPN) that observes dose volume histograms of a plan and outputs an action to adjust organ weighting factors, similar to the behaviors of a human planner. We trained the WTPN via end-to-end deep reinforcement learning. Experience replay was performed with the epsilon greedy algorithm. After training was completed, we applied the trained WTPN to guide treatment planning of five testing patient cases. It was found that the trained WTPN successfully learnt the treatment planning goals and was able to guide the weight tuning process. On average, the quality score of plans generated under the WTPN's guidance was improved by ~8.5% compared to the initial plan with arbitrarily set weights, and by 10.7% compared to the plans generated by human planners. To our knowledge, this was the first time that a tool was developed to adjust organ weights for the treatment planning optimization problem in a human-like fashion based on intelligence learnt from a training process, which was different from existing strategies based on pre-defined rules. The study demonstrated potential feasibility to develop intelligent treatment planning approaches via deep reinforcement learning.

DeepPET: A deep encoder-decoder network for directly solving the PET image reconstruction inverse problem

The purpose of this research was to implement a deep learning network to overcome two of the major bottlenecks in improved image reconstruction for clinical positron emission tomography (PET). These are the lack of an automated means for the optimization of advanced image reconstruction algorithms, and the computational expense associated with these state-of-the art methods. We thus present a novel end-to-end PET image reconstruction technique, called DeepPET, based on a deep convolutional encoder-decoder network, which takes PET sinogram data as input and directly and quickly outputs high quality, quantitative PET images. Using simulated data derived from a whole-body digital phantom, we randomly sampled the configurable parameters to generate realistic images, which were each augmented to a total of more than 291,000 reference images. Realistic PET acquisitions of these images were simulated, resulting in noisy sinogram data, used for training, validation, and testing the DeepPET network. We demonstrated that DeepPET generates higher quality images compared to conventional techniques, in terms of relative root mean squared error (11%/53% lower than ordered subset expectation maximization (OSEM)/filtered back-projection (FBP), structural similarity index (1%/11% higher than OSEM/FBP), and peak signal-to-noise ratio (1.1/3.8 dB higher than OSEM/FBP). In addition, we show that DeepPET reconstructs images 108 and 3 times faster than OSEM and FBP, respectively. Finally, DeepPET was successfully applied to real clinical data. This study shows that an end-to-end encoder-decoder network can produce high quality PET images at a fraction of the time compared to conventional methods.

Multi-energy computed tomography reconstruction using a nonlocal spectral similarity model

Multi-energy computed tomography (MECT) is able to acquire simultaneous multi-energy measurements from one scan. In addition, it allows material differentiation and quantification effectively. However, due to the limited energy bin width, the number of photons detected in an energy-specific channel is smaller than that in traditional CT, which results in image quality degradation. To address this issue, in this work, we develop a statistical iterative reconstruction algorithm to acquire high-quality MECT images and high-accuracy material-specific images. Specifically, this algorithm fully incorporates redundant self-similarities within nonlocal regions in the MECT image at one bin and rich spectral similarities among MECT images at all bins. For simplicity, the presented algorithm is referred to as 'MECT-NSS'. Moreover, an efficient optimization algorithm is developed to solve the MECT-NSS objective function. Then, a comprehensive evaluation of parameter selection for the MECT-NSS algorithm is conducted. In the experiment, the datasets include images from three phantoms and one patient to validate and evaluate the MECT-NSS reconstruction performance. The qualitative and quantitative results demonstrate that the presented MECT-NSS can successfully yield better MECT image quality and more accurate material estimation than the competing algorithms.

Statistical CT reconstruction using region-aware texture preserving regularization learning from prior normal-dose CT image

In some clinical applications, prior normal-dose CT (NdCT) images are available, and the valuable textures and structure features in them may be used to promote follow-up low-dose CT (LdCT) reconstruction. This study aims to learn texture information from the NdCT images and leverage it for follow-up LdCT image reconstruction to preserve textures and structure features. Specifically, the proposed reconstruction method first learns the texture information from those patches with similar structures in NdCT image, and the similar patches can be clustered by searching context features efficiently from the surroundings of the current patch. Then it utilizes redundant texture information from the similar patches as a priori knowledge to describe specific regions in the LdCT image. The advanced region-aware texture preserving prior is termed as 'RATP'. The main advantage of the PATP prior is that it can properly learn the texture features from available NdCT images and adaptively characterize the region-specific structures in the LdCT image. The experiments using patient data were performed to evaluate the performance of the proposed method. The proposed RATP method demonstrated superior performance in LdCT imaging compared to the filtered back projection (FBP) and statistical iterative reconstruction (SIR) methods using Gaussian regularization, Huber regularization and the original texture preserving regularization.

Robust Waveform Design of Ultrasound Arrays for Medical Imaging

Sound speed is an effective parameter in designing an optimal beamformer. In conventional ultrasound imaging systems, the beamformer is designed assuming a fixed value of speed, whereas the speed in a tissue is not known precisely and also may fluctuate by a great value. The errors in estimating sound speed may lead to a severe degradation in the reconstructed image, as mainlobe width and sidelobe level of the beampattern are sensitive to the speed variations. In this paper, we consider the design of a transmit beamformer, which is robust to the speed variations. The problem is formulated as a convex optimization problem versus the covariance matrix of the excitation waveforms to obtain a beampattern with predefined mainlobe width and a minimum sidelobe level for all possible variations of speed. Then, by eigen-analysis of the obtained covariance matrix, a set of nonidentical single-carrier short-pulses for the excitation waveforms were designed. Various simulations indicate that the proposed method can yield a robust beampattern whose mainlobe width and sidelobe level almost remain constant by 10% speed variations. In contrast, the beampatterns obtained by nonrobust methods suffer extensive changes.

Content-aware compressive magnetic resonance image reconstruction

This paper describes an adaptive approach to regularizing model-based reconstructions in magnetic resonance imaging to account for local structure or image content. In conjunction with common models like wavelet and total variation sparsity, this content-aware regularization avoids oversmoothing or compromising image features while suppressing noise and incoherent aliasing from accelerated imaging. To evaluate this regularization approach, the experiments reconstruct images from single- and multi-channel, Cartesian and non-Cartesian, brain and cardiac data. These reconstructions combine common analysis-form regularizers and autocalibrating parallel imaging (when applicable). In most cases, the results show widespread improvement in structural similarity and peak-signal-to-error ratio relative to the fully sampled images. These results suggest that this content-aware regularization can preserve local image structures such as edges while providing denoising power superior to sparsity-promoting or sparsity-reweighted regularization.