arXiv subjects

Bo Zhou

Publications and source records attributed to Bo Zhou.

At least 253 records · Page 14Linked to original sources

arXiv2021

Generation of Virtual Dual Energy Images from Standard Single-Shot Radiographs using Multi-scale and Conditional Adversarial Network

Dual-energy (DE) chest radiographs provide greater diagnostic information than standard radiographs by separating the image into bone and soft tissue, revealing suspicious lesions which may otherwise be obstructed from view. However, acquisition of DE images requires two physical scans, necessitating specialized hardware and processing, and images are prone to motion artifact. Generation of virtual DE images from standard, single-shot chest radiographs would expand the diagnostic value of standard radiographs without changing the acquisition procedure. We present a Multi-scale Conditional Adversarial Network (MCA-Net) which produces high-resolution virtual DE bone images from standard, single-shot chest radiographs. Our proposed MCA-Net is trained using the adversarial network so that it learns sharp details for the production of high-quality bone images. Then, the virtual DE soft tissue image is generated by processing the standard radiograph with the virtual bone image using a cross projection transformation. Experimental results from 210 patient DE chest radiographs demonstrated that the algorithm can produce high-quality virtual DE chest radiographs. Important structures were preserved, such as coronary calcium in bone images and lung lesions in soft tissue images. The average structure similarity index and the peak signal to noise ratio of the produced bone images in testing data were 96.4 and 41.5, which are significantly better than results from previous methods. Furthermore, our clinical evaluation results performed on the publicly available dataset indicates the clinical values of our algorithms. Thus, our algorithm can produce high-quality DE images that are potentially useful for radiologists, computer-aided diagnostics, and other diagnostic tasks.

cs.CV↗

arXiv2021

Anatomy-guided Multimodal Registration by Learning Segmentation without Ground Truth: Application to Intraprocedural CBCT/MR Liver Segmentation and Registration

Multimodal image registration has many applications in diagnostic medical imaging and image-guided interventions, such as Transcatheter Arterial Chemoembolization (TACE) of liver cancer guided by intraprocedural CBCT and pre-operative MR. The ability to register peri-procedurally acquired diagnostic images into the intraprocedural environment can potentially improve the intra-procedural tumor targeting, which will significantly improve therapeutic outcomes. However, the intra-procedural CBCT often suffers from suboptimal image quality due to lack of signal calibration for Hounsfield unit, limited FOV, and motion/metal artifacts. These non-ideal conditions make standard intensity-based multimodal registration methods infeasible to generate correct transformation across modalities. While registration based on anatomic structures, such as segmentation or landmarks, provides an efficient alternative, such anatomic structure information is not always available. One can train a deep learning-based anatomy extractor, but it requires large-scale manual annotations on specific modalities, which are often extremely time-consuming to obtain and require expert radiological readers. To tackle these issues, we leverage annotated datasets already existing in a source modality and propose an anatomy-preserving domain adaptation to segmentation network (APA2Seg-Net) for learning segmentation without target modality ground truth. The segmenters are then integrated into our anatomy-guided multimodal registration based on the robust point matching machine. Our experimental results on in-house TACE patient data demonstrated that our APA2Seg-Net can generate robust CBCT and MR liver segmentation, and the anatomy-guided registration framework with these segmenters can provide high-quality multimodal registrations. Our code is available at https://github.com/bbbbbbzhou/APA2Seg-Net.

eess.IV↗

arXiv2021

Lifelong Learning for Minimizing Age of Information in Internet of Things Networks

In this paper, a lifelong learning problem is studied for an Internet of Things (IoT) system. In the considered model, each IoT device aims to balance its information freshness and energy consumption tradeoff by controlling its computational resource allocation at each time slot under dynamic environments. An unmanned aerial vehicle (UAV) is deployed as a flying base station so as to enable the IoT devices to adapt to novel environments. To this end, a new lifelong reinforcement learning algorithm, used by the UAV, is proposed in order to adapt the operation of the devices at each visit by the UAV. By using the experience from previously visited devices and environments, the UAV can help devices adapt faster to future states of their environment. To do so, a knowledge base shared by all devices is maintained at the UAV. Simulation results show that the proposed algorithm can converge $25\%$ to $50\%$ faster than a policy gradient baseline algorithm that optimizes each device's decision making problem in isolation.

cs.NI↗

arXiv2021

Centralized and Distributed Age of Information Minimization with non-linear Aging Functions in the Internet of Things

Resource management in Internet of Things (IoT) systems is a major challenge due to the massive scale and heterogeneity of the IoT system. For instance, most IoT applications require timely delivery of collected information, which is a key challenge for the IoT. In this paper, novel centralized and distributed resource allocation schemes are proposed to enable IoT devices to share limited communication resources and to transmit IoT messages in a timely manner. In the considered system, the timeliness of information is captured using non-linear age of information (AoI) metrics that can naturally quantify the freshness of information. To model the inherent heterogeneity of the IoT system, the non-linear aging functions are defined in terms of IoT device types and message content. To minimize AoI, the proposed resource management schemes allocate the limited communication resources considering AoI. In particular, the proposed centralized scheme enables the base station to learn the device types and to determine aging functions. Moreover, the proposed distributed scheme enables the devices to share the limited communication resources based on available information on other devices and their AoI. The convergence of the proposed distributed scheme is proved, and the effectiveness in reducing the AoI with partial information is analyzed. Simulation results show that the proposed centralized scheme achieves significantly lower average instantaneous AoI when compared to simple centralized allocation without learning, while the proposed distributed scheme achieves significantly lower average instantaneous AoI when compared to random allocation. The results also show that the proposed centralized scheme outperforms the proposed distributed scheme in almost all cases, but the distributed approach is more viable for a massive IoT.

cs.IT↗

arXiv2020

Performance Analysis of Age of Information in Ultra-Dense Internet of Things (IoT) Systems with Noisy Channels

In this paper, a dense Internet of Things (IoT) monitoring system is studied in which a large number of devices contend for transmitting timely status packets to their corresponding receivers over wireless noisy channels, using a carrier sense multiple access (CSMA) scheme. When each device completes one transmission, due to possible transmission failure, two cases with and without transmission feedback to each device must be considered. Particularly, for the case with no feedback, the device uses policy (I): It will go to an idle state and release the channel regardless of the outcome of the transmission. For the case with perfect feedback, if the transmission succeeds, the device will go to an idle state, otherwise, it uses either policy (W), i.e., it will go to a waiting state and re-contend for channel access; or it uses policy (S), i.e., it will stay at a service state and occupy this channel to attempt another transmission. For those three policies, the closed-form expressions of the average age of information (AoI) of each device are characterized under schemes with and without preemption in service. It is shown that, for each policy, the scheme with preemption in service always achieves a smaller average AoI, compared with the scheme without preemption. Then, a mean-field approximation approach with guaranteed accuracy is developed to analyze the asymptotic performance for the considered system with an infinite number of devices and the effects of the system parameters on the average AoI are characterized. Simulation results show that the proposed mean-field approximation is accurate even for a small number of devices. The results also show that policy (S) achieves the smallest average AoI compared with policies (I) and (W), and the average AoI does not always decrease with the arrival rate for all three policies.

cs.IT↗

arXiv2020

Long-distance tiny face detection based on enhanced YOLOv3 for unmanned system

Remote tiny face detection applied in unmanned system is a challeng-ing work. The detector cannot obtain sufficient context semantic information due to the relatively long distance. The received poor fine-grained features make the face detection less accurate and robust. To solve the problem of long-distance detection of tiny faces, we propose an enhanced network model (YOLOv3-C) based on the YOLOv3 algorithm for unmanned platform. In this model, we bring in multi-scale features from feature pyramid networks and make the features fu-sion to adjust prediction feature map of the output, which improves the sensitivity of the entire algorithm for tiny target faces. The enhanced model improves the accuracy of tiny face detection in the cases of long-distance and high-density crowds. The experimental evaluation results demonstrated the superior perfor-mance of the proposed YOLOv3-C in comparison with other relevant detectors in remote tiny face detection. It is worth mentioning that our proposed method achieves comparable performance with the state of the art YOLOv4[1] in the tiny face detection tasks.

cs.CV↗

arXiv2020

AIM 2020 Challenge on Real Image Super-Resolution: Methods and Results

This paper introduces the real image Super-Resolution (SR) challenge that was part of the Advances in Image Manipulation (AIM) workshop, held in conjunction with ECCV 2020. This challenge involves three tracks to super-resolve an input image for $\times$2, $\times$3 and $\times$4 scaling factors, respectively. The goal is to attract more attention to realistic image degradation for the SR task, which is much more complicated and challenging, and contributes to real-world image super-resolution applications. 452 participants were registered for three tracks in total, and 24 teams submitted their results. They gauge the state-of-the-art approaches for real image SR in terms of PSNR and SSIM.

cs.CV↗

arXiv2020

Semiparametric Testing with Highly Persistent Predictors

We address the issue of semiparametric efficiency in the bivariate regression problem with a highly persistent predictor, where the joint distribution of the innovations is regarded an infinite-dimensional nuisance parameter. Using a structural representation of the limit experiment and exploiting invariance relationships therein, we construct invariant point-optimal tests for the regression coefficient of interest. This approach naturally leads to a family of feasible tests based on the component-wise ranks of the innovations that can gain considerable power relative to existing tests under non-Gaussian innovation distributions, while behaving equivalently under Gaussianity. When an i.i.d. assumption on the innovations is appropriate for the data at hand, our tests exploit the efficiency gains possible. Moreover, we show by simulation that our test remains well behaved under some forms of conditional heteroskedasticity.

econ.EM↗

arXiv2020

Simultaneous Denoising and Motion Estimation for Low-dose Gated PET using a Siamese Adversarial Network with Gate-to-Gate Consistency Learning

Gating is commonly used in PET imaging to reduce respiratory motion blurring and facilitate more sophisticated motion correction methods. In the applications of low dose PET, however, reducing injection dose causes increased noise and reduces signal-to-noise ratio (SNR), subsequently corrupting the motion estimation/correction steps, causing inferior image quality. To tackle these issues, we first propose a Siamese adversarial network (SAN) that can efficiently recover high dose gated image volume from low dose gated image volume. To ensure the appearance consistency between the recovered gated volumes, we then utilize a pre-trained motion estimation network incorporated into SAN that enables the constraint of gate-to-gate (G2G) consistency. With high-quality recovered gated volumes, gate-to-gate motion vectors can be simultaneously outputted from the motion estimation network. Comprehensive evaluations on a low dose gated PET dataset of 29 subjects demonstrate that our method can effectively recover the low dose gated PET volumes, with an average PSNR of 37.16 and SSIM of 0.97, and simultaneously generate robust motion estimation that could benefit subsequent motion corrections.

eess.IV↗

arXiv2020

Limited View Tomographic Reconstruction Using a Deep Recurrent Framework with Residual Dense Spatial-Channel Attention Network and Sinogram Consistency

Limited view tomographic reconstruction aims to reconstruct a tomographic image from a limited number of sinogram or projection views arising from sparse view or limited angle acquisitions that reduce radiation dose or shorten scanning time. However, such a reconstruction suffers from high noise and severe artifacts due to the incompleteness of sinogram. To derive quality reconstruction, previous state-of-the-art methods use UNet-like neural architectures to directly predict the full view reconstruction from limited view data; but these methods leave the deep network architecture issue largely intact and cannot guarantee the consistency between the sinogram of the reconstructed image and the acquired sinogram, leading to a non-ideal reconstruction. In this work, we propose a novel recurrent reconstruction framework that stacks the same block multiple times. The recurrent block consists of a custom-designed residual dense spatial-channel attention network. Further, we develop a sinogram consistency layer interleaved in our recurrent framework in order to ensure that the sampled sinogram is consistent with the sinogram of the intermediate outputs of the recurrent blocks. We evaluate our methods on two datasets. Our experimental results on AAPM Low Dose CT Grand Challenge datasets demonstrate that our algorithm achieves a consistent and significant improvement over the existing state-of-the-art neural methods on both limited angle reconstruction (over 5dB better in terms of PSNR) and sparse view reconstruction (about 4dB better in term of PSNR). In addition, our experimental results on Deep Lesion datasets demonstrate that our method is able to generate high-quality reconstruction for 8 major lesion types.

eess.IV↗

arXiv2020

On extremal leaf status and internal status of trees

For a vertex $u$ of a tree $T$, the leaf (internal, respectively) status of $u$ is the sum of the distances from $u$ to all leaves (internal vertices, respectively) of $T$. The minimum (maximum, respectively) leaf status of a tree $T$ is the minimum (maximum, respectively) leaf statuses of all vertices of $T$. The minimum (maximum, respectively) internal status of a tree $T$ is the minimum (maximum, respectively) internal statuses of all vertices of $T$. We give the smallest and largest values for the minimum leaf status, maximum leaf status, minimum internal status, and maximum internal status of a tree and characterize the extremal cases. We also discuss these parameters of a tree with given diameter or maximum degree.

cs.DM↗

arXiv2020

On the Age of Information in Internet of Things Systems with Correlated Devices

In this paper, a real-time Internet of Things (IoT) monitoring system is considered in which multiple IoT devices must transmit timely updates on the status information of a common underlying physical process to a common destination. In particular, a real-world IoT scenario is considered in which multiple (partially) observed status information by different IoT devices are required at the destination, so that the real-time status of the physical process can be properly re-constructed. By taking into account such correlated status information at the IoT devices, the problem of IoT device scheduling is studied in order to jointly minimize the average age of information (AoI) at the destination and the average energy cost at the IoT devices. Particularly, two types of IoT devices are considered: Type-I devices whose status updates randomly arrive and type-II devices whose status updates can be generated-at-will with an associated sampling cost. This stochastic problem is formulated as an infinite horizon average cost Markov decision process (MDP). The optimal scheduling policy is shown to be threshold-based with respect to the AoI at the destination, and the threshold is non-increasing with the channel condition of each device. For a special case in which all devices are type-II, the original MDP can be reduced to an MDP with much smaller state and action spaces. The optimal policy is further shown to have a similar threshold-based structure and the threshold is non-decreasing with an energy cost function of the devices. Simulation results illustrate the structure of the optimal policy and show the effectiveness of the optimal policy compared with a myopic baseline policy.

cs.IT↗

arXiv2020

DuDoRNet: Learning a Dual-Domain Recurrent Network for Fast MRI Reconstruction with Deep T1 Prior

MRI with multiple protocols is commonly used for diagnosis, but it suffers from a long acquisition time, which yields the image quality vulnerable to say motion artifacts. To accelerate, various methods have been proposed to reconstruct full images from under-sampled k-space data. However, these algorithms are inadequate for two main reasons. Firstly, aliasing artifacts generated in the image domain are structural and non-local, so that sole image domain restoration is insufficient. Secondly, though MRI comprises multiple protocols during one exam, almost all previous studies only employ the reconstruction of an individual protocol using a highly distorted undersampled image as input, leaving the use of fully-sampled short protocol (say T1) as complementary information highly underexplored. In this work, we address the above two limitations by proposing a Dual Domain Recurrent Network (DuDoRNet) with deep T1 prior embedded to simultaneously recover k-space and images for accelerating the acquisition of MRI with a long imaging protocol. Specifically, a Dilated Residual Dense Network (DRDNet) is customized for dual domain restorations from undersampled MRI data. Extensive experiments on different sampling patterns and acceleration rates demonstrate that our method consistently outperforms state-of-the-art methods, and can reconstruct high-quality MRI.

eess.IV↗

arXiv2020

A deep learning-facilitated radiomics solution for the prediction of lung lesion shrinkage in non-small cell lung cancer trials

Herein we propose a deep learning-based approach for the prediction of lung lesion response based on radiomic features extracted from clinical CT scans of patients in non-small cell lung cancer trials. The approach starts with the classification of lung lesions from the set of primary and metastatic lesions at various anatomic locations. Focusing on the lung lesions, we perform automatic segmentation to extract their 3D volumes. Radiomic features are then extracted from the lesion on the pre-treatment scan and the first follow-up scan to predict which lesions will shrink at least 30% in diameter during treatment (either Pembrolizumab or combinations of chemotherapy and Pembrolizumab), which is defined as a partial response by the Response Evaluation Criteria In Solid Tumors (RECIST) guidelines. A 5-fold cross validation on the training set led to an AUC of 0.84 +/- 0.03, and the prediction on the testing dataset reached AUC of 0.73 +/- 0.02 for the outcome of 30% diameter shrinkage.

eess.IV↗

arXiv2020

Risk-Aware Optimization of Age of Information in the Internet of Things

Minimization of the expected value of age of information (AoI) is a risk-neutral approach, and it thus cannot capture rare, yet critical, events with potentially large AoI. In order to capture the effect of these events, in this paper, the notion of conditional value-at-risk (CVaR) is proposed as an effective coherent risk measure that is suitable for minimization of AoI for real-time IoT status updates. In the considered monitoring system, an IoT device monitors a physical process and sends the status updates to a remote receiver with an updating cost. The optimal status update process is designed to jointly minimize the AoI at the receiver, the CVaR of the AoI at the receiver, and the energy cost. This stochastic optimization problem is formulated as an infinite horizon discounted risk-aware Markov decision process (MDP), which is computationally intractable due to the time inconsistency of the CVaR. By exploiting the special properties of coherent risk measures, the risk-aware MDP is reduced to a standard MDP with an augmented state space, for which we derive the optimal stationary policy using dynamic programming. In particular, the optimal history-dependent policy of the risk-aware MDP is shown to depend on the history only through the augmented system states and can be readily constructed using the optimal stationary policy of the augmented MDP. The proposed solution is shown to be computationally tractable and able to minimize the AoI in real-time IoT monitoring systems in a risk-aware manner.

cs.IT↗

arXiv2020

Age of Information Analysis for Dynamic Spectrum Sharing

Timely information updates are critical to time-sensitive applications in networked monitoring and control systems. In this paper, the problem of real-time status update is considered for a cognitive radio network (CRN), in which the secondary user (SU) can relay the status packets from the primary user (PU) to the destination. In the considered CRN, the SU has opportunities to access the spectrum owned by the PU to send its own status packets to the destination. The freshness of information is measured by the age of information (AoI) metric. The problem of minimizing the average AoI and energy consumption by developing new optimal status update and packet relaying schemes for the SU is addressed under an average AoI constraint for the PU. This problem is formulated as a constrained Markov decision process (CMDP). The monotonic and decomposable properties of the value function are characterized and then used to show that the optimal update and relaying policy is threshold-based with respect to the AoI of the SU. These structures reveal a tradeoff between the AoI of the SU and the energy consumption as well as between the AoI of the SU and the AoI of the PU. An asymptotically optimal algorithm is proposed. Numerical results are then used to show the effectiveness of the proposed policy.

cs.NI↗

arXiv2020

Minimum Age of Information in the Internet of Things with Non-uniform Status Packet Sizes

In this paper, a real-time Internet of Things (IoT) monitoring system is considered in which the IoT devices are scheduled to sample underlying physical processes and send the status updates to a common destination. In a real-world IoT, due to the possibly different dynamics of each physical process, the sizes of the status updates for different devices are often different and each status update typically requires multiple transmission slots. By taking into account such multi-time slot transmissions with non-uniform sizes of the status updates under noisy channels, the problem of joint device scheduling and status sampling is studied in order to minimize the average age of information (AoI) at the destination. This stochastic problem is formulated as an infinite horizon average cost Markov decision process (MDP). The monotonicity of the value function of the MDP is characterized and then used to show that the optimal scheduling and sampling policy is threshold-based with respect to the AoI at each device. To overcome the curse of dimensionality, a low-complexity suboptimal policy is proposed through a semi-randomized base policy and linear approximated value functions. The proposed suboptimal policy is shown to exhibit a similar structure to the optimal policy, which provides a structural base for its effective performance. A structure-aware algorithm is then developed to obtain the suboptimal policy. The analytical results are further extended to the IoT monitoring system with random status update arrivals, for which, the optimal scheduling and sampling policy is also shown to be threshold-based with the AoI at each device. Simulation results illustrate the structures of the optimal policy and show a near-optimal AoI performance resulting from the proposed suboptimal solution approach.

cs.IT↗

arXiv2019

Efficient and Robust Reinforcement Learning with Uncertainty-based Value Expansion

By integrating dynamics models into model-free reinforcement learning (RL) methods, model-based value expansion (MVE) algorithms have shown a significant advantage in sample efficiency as well as value estimation. However, these methods suffer from higher function approximation errors than model-free methods in stochastic environments due to a lack of modeling the environmental randomness. As a result, their performance lags behind the best model-free algorithms in some challenging scenarios. In this paper, we propose a novel Hybrid-RL method that builds on MVE, namely the Risk Averse Value Expansion (RAVE). With imaginative rollouts generated by an ensemble of probabilistic dynamics models, we further introduce the aversion of risks by seeking the lower confidence bound of the estimation. Experiments on a range of challenging environments show that by modeling the uncertainty completely, RAVE substantially enhances the robustness of previous model-based methods, and yields state-of-the-art performance. With this technique, our solution gets the first place in NeurIPS 2019: Learn to Move.

cs.LG↗