arXiv subjects

Ling Zhang

Publications and source records attributed to Ling Zhang.

At least 127 records · Page 7Linked to original sources

arXiv2019

Spatio-Temporal Convolutional LSTMs for Tumor Growth Prediction by Learning 4D Longitudinal Patient Data

Prognostic tumor growth modeling via volumetric medical imaging observations can potentially lead to better outcomes of tumor treatment and surgical planning. Recent advances of convolutional networks have demonstrated higher accuracy than traditional mathematical models in predicting future tumor volumes. This indicates that deep learning-based techniques may have great potentials on addressing such problem. However, current 2D patch-based modeling approaches cannot make full use of the spatio-temporal imaging context of the tumor's longitudinal 4D (3D + time) data. Moreover, they are incapable to predict clinically-relevant tumor properties, other than volumes. In this paper, we exploit to formulate the tumor growth process through convolutional Long Short-Term Memory (ConvLSTM) that extract tumor's static imaging appearances and capture its temporal dynamic changes within a single network. We extend ConvLSTM into the spatio-temporal domain (ST-ConvLSTM) by jointly learning the inter-slice 3D contexts and the longitudinal or temporal dynamics from multiple patient studies. Our approach can incorporate other non-imaging patient information in an end-to-end trainable manner. Experiments are conducted on the largest 4D longitudinal tumor dataset of 33 patients to date. Results validate that the ST-ConvLSTM produces a Dice score of 83.2%+-5.1% and a RVD of 11.2%+-10.8%, both significantly outperforming (p<0.05) other compared methods of linear model, ConvLSTM, and generative adversarial network (GAN) under the metric of predicting future tumor volumes. Additionally, our new method enables the prediction of both cell density and CT intensity numbers. Last, we demonstrate the generalizability of ST-ConvLSTM by employing it in 4D medical image segmentation task, which achieves an averaged Dice score of 86.3+-1.2% for left-ventricle segmentation in 4D ultrasound with 3 seconds per patient.

cs.CV↗

arXiv2019

Unsupervised Feature Learning for Point Cloud by Contrasting and Clustering With Graph Convolutional Neural Network

To alleviate the cost of collecting and annotating large-scale point cloud datasets, we propose an unsupervised learning approach to learn features from unlabeled point cloud "3D object" dataset by using part contrasting and object clustering with deep graph neural networks (GNNs). In the contrast learning step, all the samples in the 3D object dataset are cut into two parts and put into a "part" dataset. Then a contrast learning GNN (ContrastNet) is trained to verify whether two randomly sampled parts from the part dataset belong to the same object. In the cluster learning step, the trained ContrastNet is applied to all the samples in the original 3D object dataset to extract features, which are used to group the samples into clusters. Then another GNN for clustering learning (ClusterNet) is trained to predict the cluster ID of all the training samples. The contrasting learning forces the ContrastNet to learn high-level semantic features of objects but probably ignores low-level features, while the ClusterNet improves the quality of learned features by being trained to discover objects that probably belong to the same semantic categories by the use of cluster IDs. We have conducted extensive experiments to evaluate the proposed framework on point cloud classification tasks. The proposed unsupervised learning approach obtained comparable performance to the state-of-the-art unsupervised learning methods that used much more complicated network structures. The code of this work is publicly available via: https://github.com/lingzhang1/ContrastNet.

cs.CV↗

arXiv2019

ARGAN: Attentive Recurrent Generative Adversarial Network for Shadow Detection and Removal

In this paper we propose an attentive recurrent generative adversarial network (ARGAN) to detect and remove shadows in an image. The generator consists of multiple progressive steps. At each step a shadow attention detector is firstly exploited to generate an attention map which specifies shadow regions in the input image.Given the attention map, a negative residual by a shadow remover encoder will recover a shadow-lighter or even a shadow-free image. A discriminator is designed to classify whether the output image in the last progressive step is real or fake. Moreover, ARGAN is suitable to be trained with a semi-supervised strategy to make full use of sufficient unsupervised data. The experiments on four public datasets have demonstrated that our ARGAN is robust to detect both simple and complex shadows and to produce more realistic shadow removal results. It outperforms the state-of-the-art methods, especially in detail of recovering shadow areas.

cs.CV↗

arXiv2019

Correlation via synthesis: end-to-end nodule image generation and radiogenomic map learning based on generative adversarial network

Radiogenomic map linking image features and gene expression profiles is useful for noninvasively identifying molecular properties of a particular type of disease. Conventionally, such map is produced in three separate steps: 1) gene-clustering to "metagenes", 2) image feature extraction, and 3) statistical correlation between metagenes and image features. Each step is independently performed and relies on arbitrary measurements. In this work, we investigate the potential of an end-to-end method fusing gene data with image features to generate synthetic image and learn radiogenomic map simultaneously. To achieve this goal, we develop a generative adversarial network (GAN) conditioned on both background images and gene expression profiles, synthesizing the corresponding image. Image and gene features are fused at different scales to ensure the realism and quality of the synthesized image. We tested our method on non-small cell lung cancer (NSCLC) dataset. Results demonstrate that the proposed method produces realistic synthetic images, and provides a promising way to find gene-image relationship in a holistic end-to-end manner.

cs.CV↗

arXiv2019

Scenario Forecasting of Residential Load Profiles

Load forecasting is an integral part of power system operations and planning. Due to the increasing penetration of rooftop PV, electric vehicles and demand response applications, forecasting the load of individual and a small group of households is becoming increasingly important. Forecasting the load accurately, however, is considerable more difficult when only a few households are included. A way to mitigate this challenge is to provide a set of scenarios instead of one point forecast, so an operator or utility can consider a range of behaviors. This paper proposes a novel scenario forecasting approach for residential load using flow-based conditional generative models. Compared to existing scenario forecasting methods, our approach can generate scenarios that are not only able to infer possible future realizations of residential load from the observed historical data but also realistic enough to cover a wide range of behaviors. Particularly, the flow-based models utilize a flow of reversible transformations to maximize the value of conditional density function of future load given the past observations. In order to better capture the complex temporal dependency of the forecasted future load on the condition, we extend the structure design for the reversible transformations in flow-based conditional generative models by strengthening the coupling between the output and the conditional input in the transformations. The simulation results show the flow-based designs outperform existing methods in scenario forecasting for residential load by both providing more accurate and more diverse scenarios.

eess.SY↗

arXiv2019

Vulnerabilities of Power System Operations to Load Forecasting Data Injection Attacks

We study the security threats of power system operation brought by a class of data injection attacks upon load forecasting algorithms. In particular, with minimal assumptions on the knowledge and ability of the attacker, we design attack data on input features for load forecasting algorithms in a black-box approach. System operators can be oblivious of such wrong load forecasts, which lead to uneconomical or even insecure decisions in commitment and dispatch. This paper is the first attempt to bring up the security issues of load forecasting algorithms to our knowledge, and show that accurate load forecasting algorithm is not necessarily robust to malicious attacks. More severely, attackers are able to design targeted attacks on system operations strategically with additional topology information. We demonstrate the impact of load forecasting attacks on two IEEE test cases. We show our attack strategy is able to cause load shedding with high probability under various settings in the 14-bus test case, and also demonstrate system-wide threats in the 118-bus test case with limited local attacks.

math.OC↗

arXiv2019

When Unseen Domain Generalization is Unnecessary? Rethinking Data Augmentation

Recent advances in deep learning for medical image segmentation demonstrate expert-level accuracy. However, in clinically realistic environments, such methods have marginal performance due to differences in image domains, including different imaging protocols, device vendors and patient populations. Here we consider the problem of domain generalization, when a model is trained once, and its performance generalizes to unseen domains. Intuitively, within a specific medical imaging modality the domain differences are smaller relative to natural images domain variability. We rethink data augmentation for medical 3D images and propose a deep stacked transformations (DST) approach for domain generalization. Specifically, a series of n stacked transformations are applied to each image in each mini-batch during network training to account for the contribution of domain-specific shifts in medical images. We comprehensively evaluate our method on three tasks: segmentation of whole prostate from 3D MRI, left atrial from 3D MRI, and left ventricle from 3D ultrasound. We demonstrate that when trained on a small source dataset, (i) on average, DST models on unseen datasets degrade only by 11% (Dice score change), compared to the conventional augmentation (degrading 39%) and CycleGAN-based domain adaptation method (degrading 25%); (ii) when evaluation on the same domain, DST is also better albeit only marginally. (iii) When training on large-sized data, DST on unseen domains reaches performance of state-of-the-art fully supervised models. These findings establish a strong benchmark for the study of domain generalization in medical imaging, and can be generalized to the design of robust deep segmentation models for clinical deployment.

cs.CV↗

arXiv2019

Non-Majorana Origin of the Half-Quantized Conductance Plateau in Quantum Anomalous Hall Insulator and Superconductor Hybrid Structures

A quantum anomalous Hall (QAH) insulator coupled to an s-wave superconductor is predicted to harbor a topological superconducting phase, the elementary excitations of which (i.e. Majorana fermions) can form topological qubits upon non-Abelian braiding operations. A recent transport experiment interprets the half-quantized two-terminal conductance plateau as the presence of chiral Majorana fermions in a millimeter-size QAH-Nb hybrid structure. However, there are concerns about this interpretation because non-Majorana mechanisms can also generate similar signatures, especially in a disordered QAH system. Here, we fabricated QAH-Nb hybrid structures and studied the QAH-Nb contact transparency and its effect on the corresponding two-terminal conductance. When the QAH film is tuned to the metallic regime by electric gating, we observed a sharp zero-bias enhancement in the differential conductance, up to 80% at zero magnetic field. This large enhancement suggests high probability of Andreev reflection and transparent interface between the magnetic topological insulator (TI) and Nb layers. When the magnetic TI film is in the QAH state with well-aligned magnetization, we found that the two-terminal conductance is always half-quantized. Our experiment provides a comprehensive understanding of the superconducting proximity effect observed in QAH-superconductor hybrid structures and shows that the half-quantized conductance plateau is unlikely to be induced by chiral Majorana fermions.

cond-mat.mes-hall↗

arXiv2019

Clustering Based Hybrid Precoding Design for Multi-User Massive MIMO Systems

Hybrid precoding has been recognized as a promising technology to combat the path loss of millimeter wave signals in massive multiple-input multiple-output (MIMO) systems. However, due to the joint optimization of the digital and analog precoding matrices as well as extra constraints for the analog part, the hybrid precoding design is still a tough issue in current research. In this paper, we adopt the thought of clustering in unsupervised learning and provide design schemes for fully-connected hybrid precoding (FHP) and adaptively-connected hybrid precoding (AHP) in multi-user massive MIMO systems. For FHP, we propose the hierarchical-agglomerative-clustering-based (HAC-based) scheme to explore the relevance among RF chains in optimal hybrid procoding design. The similar RF chains are merged into an individual RF chain when insufficient RF chains are equipped at the base station (BS). For AHP, we propose the modified-K-means-based (MKM-based) scheme to explore the relevance among antennas at the BS. The similar antennas are supported by the same RF chain to make full use of the flexible connection in AHP. Particularly, in proposed MKM-based AHP design, the clustering centers are updated by alternating-optimum-based (AO-based) scheme with a special initialization method, which is capable to individually provide feasible sub-connected hybrid precoding (SHP) design. Simulation results highlight the superior spectrum efficiency of proposed HAC-based FHP scheme, and the high power efficiency of proposed MKM-based AHP scheme. Moreover, all the proposed schemes are clarified to effectively handle the inter-user interference and outperform the existing work.

eess.SP↗

arXiv2019

Observation of Interfacial Antiferromagnetic Coupling between Magnetic Topological Insulator and Antiferromagnetic Insulator

Inducing magnetic orders in a topological insulator (TI) to break its time reversal symmetry has been predicted to reveal many exotic topological quantum phenomena. The manipulation of magnetic orders in a TI layer can play a key role in harnessing these quantum phenomena towards technological applications. Here we fabricated a thin magnetic TI film on an antiferromagnetic (AFM) insulator Cr2O3 layer and found that the magnetic moments of the magnetic TI layer and the surface spins of the Cr2O3 layers favor interfacial AFM coupling. Field cooling studies show a crossover from negative to positive exchange bias clarifying the competition between the interfacial AFM coupling energy and the Zeeman energy in the AFM insulator layer. The interfacial exchange coupling also enhances the Curie temperature of the magnetic TI layer. The unique interfacial AFM alignment in magnetic TI on AFM insulator heterostructures opens a new route toward manipulating the interplay between topological states and magnetic orders in spin-engineered heterostructures, facilitating the exploration of proof-of-concept TI-based spintronic and electronic devices with multi-functionality and low power consumption.

cond-mat.mtrl-sci↗

arXiv2018

Solving Poisson's Equation using Deep Learning in Particle Simulation of PN Junction

Simulating the dynamic characteristics of a PN junction at the microscopic level requires solving the Poisson's equation at every time step. Solving at every time step is a necessary but time-consuming process when using the traditional finite difference (FDM) approach. Deep learning is a powerful technique to fit complex functions. In this work, deep learning is utilized to accelerate solving Poisson's equation in a PN junction. The role of the boundary condition is emphasized in the loss function to ensure a better fitting. The resulting I-V curve for the PN junction, using the deep learning solver presented in this work, shows a perfect match to the I-V curve obtained using the finite difference method, with the advantage of being 10 times faster at every time step.

physics.comp-ph↗

arXiv2018

Fine-Grained Classification of Cervical Cells Using Morphological and Appearance Based Convolutional Neural Networks

Fine-grained classification of cervical cells into different abnormality levels is of great clinical importance but remains very challenging. Contrary to traditional classification methods that rely on hand-crafted or engineered features, convolution neural network (CNN) can classify cervical cells based on automatically learned deep features. However, CNN in previous studies do not involve cell morphological information, and it is unknown whether morphological features can be directly modeled by CNN to classify cervical cells. This paper presents a CNN-based method that combines cell image appearance with cell morphology for classification of cervical cells in Pap smear. The training cervical cell dataset consists of adaptively re-sampled image patches coarsely centered on the nuclei. Several CNN models (AlexNet, GoogleNet, ResNet and DenseNet) pre-trained on ImageNet dataset were fine-tuned on the cervical dataset for comparison. The proposed method is evaluated on the Herlev cervical dataset by five-fold cross-validation at patient level splitting. Results show that by adding cytoplasm and nucleus masks as raw morphological information into appearance-based CNN learning, higher classification accuracies can be achieved in general. Among the four CNN models, GoogleNet fed with both morphological and appearance information obtains the highest classification accuracies of 94.5% for 2-class classification task and 64.5% for 7-class classification task. Our method demonstrates that combining cervical cell morphology with appearance information can provide improved classification performance, which is clinically important for early diagnosis of cervical dysplastic changes.

cs.CV↗

arXiv2018

Riemann-Hilbert approach for a mixed coupled nonlinear Schrödinger system and its soliton solutions

In this work, we examine the integrable mixed coupled nonlinear Schrödinger (mCNLS) system, which describe the propagation of an optical pulse in a birefringent optical fiber. By the Riemann-Hilbert(RH) approach, the N-soliton solutions of the mCNLS system can be expressed explicitly when the jump matrix of a specific RH problem is a $3\times3$ unit matrix. As a special example, the expression of one- and two-soliton are displayed explicitly. More generally, as a promotion, an integrable generalized multi-component NLS system with its linear spectral problem be discussed. It is hoped that our results can help enrich the nonlinear dynamical behaviors of the mCNLS.

math.AP↗

arXiv2018

Riemann-Hilbert method and soliton solutions in the system of two-component Hirota equations

In this letter we examine the two-component Hirota (TH) equations which describes the pulse propagation in a coupled fiber with higher-order dispersion and self-steepening. As the TH equations is a complete integrable system, which admits a $3\times 3$ Ablowitz-Kaup-Newell-Segu(AKNS)-type Lax pair, we obtain the general N-soliton solutions of the TH equations via the Riemann-Hilbert(RH) method when the jump matrix of a specific RH problem is a $3\times3$ unit matrix. As an example, the expression of one- and two-soliton are displayed explicitly.

math.AP↗

arXiv2018

Deep Lesion Graphs in the Wild: Relationship Learning and Organization of Significant Radiology Image Findings in a Diverse Large-scale Lesion Database

Radiologists in their daily work routinely find and annotate significant abnormalities on a large number of radiology images. Such abnormalities, or lesions, have collected over years and stored in hospitals' picture archiving and communication systems. However, they are basically unsorted and lack semantic annotations like type and location. In this paper, we aim to organize and explore them by learning a deep feature representation for each lesion. A large-scale and comprehensive dataset, DeepLesion, is introduced for this task. DeepLesion contains bounding boxes and size measurements of over 32K lesions. To model their similarity relationship, we leverage multiple supervision information including types, self-supervised location coordinates and sizes. They require little manual annotation effort but describe useful attributes of the lesions. Then, a triplet network is utilized to learn lesion embeddings with a sequential sampling strategy to depict their hierarchical similarity structure. Experiments show promising qualitative and quantitative results on lesion retrieval, clustering, and classification. The learned embeddings can be further employed to build a lesion graph for various clinically useful applications. We propose algorithms for intra-patient lesion matching and missing annotation mining. Experimental results validate their effectiveness.

cs.CV↗

arXiv2018

Self-Learning to Detect and Segment Cysts in Lung CT Images without Manual Annotation

Image segmentation is a fundamental problem in medical image analysis. In recent years, deep neural networks achieve impressive performances on many medical image segmentation tasks by supervised learning on large manually annotated data. However, expert annotations on big medical datasets are tedious, expensive or sometimes unavailable. Weakly supervised learning could reduce the effort for annotation but still required certain amounts of expertise. Recently, deep learning shows a potential to produce more accurate predictions than the original erroneous labels. Inspired by this, we introduce a very weakly supervised learning method, for cystic lesion detection and segmentation in lung CT images, without any manual annotation. Our method works in a self-learning manner, where segmentation generated in previous steps (first by unsupervised segmentation then by neural networks) is used as ground truth for the next level of network learning. Experiments on a cystic lung lesion dataset show that the deep learning could perform better than the initial unsupervised annotation, and progressively improve itself after self-learning.

cs.CV↗

arXiv2018

Deep LOGISMOS: Deep Learning Graph-based 3D Segmentation of Pancreatic Tumors on CT scans

This paper reports Deep LOGISMOS approach to 3D tumor segmentation by incorporating boundary information derived from deep contextual learning to LOGISMOS - layered optimal graph image segmentation of multiple objects and surfaces. Accurate and reliable tumor segmentation is essential to tumor growth analysis and treatment selection. A fully convolutional network (FCN), UNet, is first trained using three adjacent 2D patches centered at the tumor, providing contextual UNet segmentation and probability map for each 2D patch. The UNet segmentation is then refined by Gaussian Mixture Model (GMM) and morphological operations. The refined UNet segmentation is used to provide the initial shape boundary to build a segmentation graph. The cost for each node of the graph is determined by the UNet probability maps. Finally, a max-flow algorithm is employed to find the globally optimal solution thus obtaining the final segmentation. For evaluation, we applied the method to pancreatic tumor segmentation on a dataset of 51 CT scans, among which 30 scans were used for training and 21 for testing. With Deep LOGISMOS, DICE Similarity Coefficient (DSC) and Relative Volume Difference (RVD) reached 83.2+-7.8% and 18.6+-17.4% respectively, both are significantly improved (p<0.05) compared with contextual UNet and/or LOGISMOS alone.

cs.CV↗

arXiv2018

DeepPap: Deep Convolutional Networks for Cervical Cell Classification

Automation-assisted cervical screening via Pap smear or liquid-based cytology (LBC) is a highly effective cell imaging based cancer detection tool, where cells are partitioned into "abnormal" and "normal" categories. However, the success of most traditional classification methods relies on the presence of accurate cell segmentations. Despite sixty years of research in this field, accurate segmentation remains a challenge in the presence of cell clusters and pathologies. Moreover, previous classification methods are only built upon the extraction of hand-crafted features, such as morphology and texture. This paper addresses these limitations by proposing a method to directly classify cervical cells - without prior segmentation - based on deep features, using convolutional neural networks (ConvNets). First, the ConvNet is pre-trained on a natural image dataset. It is subsequently fine-tuned on a cervical cell dataset consisting of adaptively re-sampled image patches coarsely centered on the nuclei. In the testing phase, aggregation is used to average the prediction scores of a similar set of image patches. The proposed method is evaluated on both Pap smear and LBC datasets. Results show that our method outperforms previous algorithms in classification accuracy (98.3%), area under the curve (AUC) (0.99) values, and especially specificity (98.3%), when applied to the Herlev benchmark Pap smear dataset and evaluated using five-fold cross-validation. Similar superior performances are also achieved on the HEMLBC (H&E stained manual LBC) dataset. Our method is promising for the development of automation-assisted reading systems in primary cervical screening.

cs.CV↗