scholarly journals Connected-UNets: a deep learning architecture for breast mass segmentation

2021 ◽  
Vol 7 (1) ◽  
Author(s):  
Asma Baccouche ◽  
Begonya Garcia-Zapirain ◽  
Cristian Castillo Olea ◽  
Adel S. Elmaghraby
Keyword(s):  
Deep Learning ◽  
Network Architecture ◽  
Breast Imaging ◽  
Rapid Development ◽  
Synthetic Data ◽  
Screening Mammography ◽  
Medical Image Segmentation ◽  
Breast Mass ◽  
Generative Adversarial Network ◽  
Mass Segmentation

AbstractBreast cancer analysis implies that radiologists inspect mammograms to detect suspicious breast lesions and identify mass tumors. Artificial intelligence techniques offer automatic systems for breast mass segmentation to assist radiologists in their diagnosis. With the rapid development of deep learning and its application to medical imaging challenges, UNet and its variations is one of the state-of-the-art models for medical image segmentation that showed promising performance on mammography. In this paper, we propose an architecture, called Connected-UNets, which connects two UNets using additional modified skip connections. We integrate Atrous Spatial Pyramid Pooling (ASPP) in the two standard UNets to emphasize the contextual information within the encoder–decoder network architecture. We also apply the proposed architecture on the Attention UNet (AUNet) and the Residual UNet (ResUNet). We evaluated the proposed architectures on two publically available datasets, the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) and INbreast, and additionally on a private dataset. Experiments were also conducted using additional synthetic data using the cycle-consistent Generative Adversarial Network (CycleGAN) model between two unpaired datasets to augment and enhance the images. Qualitative and quantitative results show that the proposed architecture can achieve better automatic mass segmentation with a high Dice score of 89.52%, 95.28%, and 95.88% and Intersection over Union (IoU) score of 80.02%, 91.03%, and 92.27%, respectively, on CBIS-DDSM, INbreast, and the private dataset.

scholarly journals Generative Adversarial Networks-Based Semi-Supervised Automatic Modulation Recognition for Cognitive Radio Networks

Sensors ◽  
2018 ◽  
Vol 18 (11) ◽  
pp. 3913 ◽  
Author(s):  
Mingxuan Li ◽  
Ou Li ◽  
Guangyi Liu ◽  
Ce Zhang
Keyword(s):  
Deep Learning ◽  
Cognitive Radio ◽  
Supervised Learning ◽  
Rapid Development ◽  
Generative Adversarial Networks ◽  
Modulation Recognition ◽  
Learning Methods ◽  
Generative Adversarial Network ◽  
Adversarial Network ◽  
Automatic Modulation Recognition

With the recently explosive growth of deep learning, automatic modulation recognition has undergone rapid development. Most of the newly proposed methods are dependent on large numbers of labeled samples. We are committed to using fewer labeled samples to perform automatic modulation recognition in the cognitive radio domain. Here, a semi-supervised learning method based on adversarial training is proposed which is called signal classifier generative adversarial network. Most of the prior methods based on this technology involve computer vision applications. However, we improve the existing network structure of a generative adversarial network by adding the encoder network and a signal spatial transform module, allowing our framework to address radio signal processing tasks more efficiently. These two technical improvements effectively avoid nonconvergence and mode collapse problems caused by the complexity of the radio signals. The results of simulations show that compared with well-known deep learning methods, our method improves the classification accuracy on a synthetic radio frequency dataset by 0.1% to 12%. In addition, we verify the advantages of our method in a semi-supervised scenario and obtain a significant increase in accuracy compared with traditional semi-supervised learning methods.

Super-Resolution Based on Generative Adversarial Network for HRTEM Images

2021 ◽  
pp. 2154027
Author(s):  
Fuqi Mao ◽  
Xiaohan Guan ◽  
Ruoyu Wang ◽  
Wen Yue
Keyword(s):  
Deep Learning ◽  
High Resolution ◽  
Information Structure ◽  
Rapid Development ◽  
Super Resolution ◽  
Training Data ◽  
Learning Technology ◽  
Hrtem Image ◽  
Generative Adversarial Network ◽  
Dual Regression

As an important tool to study the microstructure and properties of materials, High Resolution Transmission Electron Microscope (HRTEM) images can obtain the lattice fringe image (reflecting the crystal plane spacing information), structure image and individual atom image (which reflects the configuration of atoms or atomic groups in crystal structure). Despite the rapid development of HTTEM devices, HRTEM images still have limited achievable resolution for human visual system. With the rapid development of deep learning technology in recent years, researchers are actively exploring the Super-resolution (SR) model based on deep learning, and the model has reached the current best level in various SR benchmarks. Using SR to reconstruct high-resolution HRTEM image is helpful to the material science research. However, there is one core issue that has not been resolved: most of these super-resolution methods require the training data to exist in pairs. In actual scenarios, especially for HRTEM images, there are no corresponding HR images. To reconstruct high quality HRTEM image, a novel Super-Resolution architecture for HRTEM images is proposed in this paper. Borrowing the idea from Dual Regression Networks (DRN), we introduce an additional dual regression structure to ESRGAN, by training the model with unpaired HRTEM images and paired nature images. Results of extensive benchmark experiments demonstrate that the proposed method achieves better performance than the most resent SISR methods with both quantitative and visual results.

scholarly journals Deep-learning Reconstruction of Three-dimensional Galaxy Distributions with Intensity Mapping Observations

2021 ◽  
Vol 923 (1) ◽  
pp. L7
Author(s):  
Kana Moriwaki ◽  
Naoki Yoshida
Keyword(s):  
Deep Learning ◽  
Emission Line ◽  
Network Architecture ◽  
Large Scale ◽  
Three Dimensional ◽  
Line Emission ◽  
Correlation Coefficients ◽  
Emission Lines ◽  
Generative Adversarial Network ◽  
Intensity Mapping

Abstract Line-intensity mapping is emerging as a novel method that can measure the collective intensity fluctuations of atomic/molecular line emission from distant galaxies. Several observational programs with various wavelengths are ongoing and planned, but there remains a critical problem of line confusion; emission lines originating from galaxies at different redshifts are confused at the same observed wavelength. We devise a generative adversarial network that extracts designated emission-line signals from noisy three-dimensional data. Our novel network architecture allows two input data, in which the same underlying large-scale structure is traced by two emission lines of H α and [Oiii], so that the network learns the relative contributions at each wavelength and is trained to decompose the respective signals. After being trained with a large number of realistic mock catalogs, the network is able to reconstruct the three-dimensional distribution of emission-line galaxies at z = 1.3−2.4. Bright galaxies are identified with a precision of 84%, and the cross correlation coefficients between the true and reconstructed intensity maps are as high as 0.8. Our deep-learning method can be readily applied to data from planned spaceborne and ground-based experiments.

scholarly journals Advancing Medical Imaging Informatics by Deep Learning-Based Domain Adaptation

2020 ◽  
Vol 29 (01) ◽  
pp. 129-138 ◽  
Author(s):  
Anirudh Choudhary ◽  
Li Tong ◽  
Yuanda Zhu ◽  
May D. Wang
Keyword(s):  
Deep Learning ◽  
Medical Imaging ◽  
Large Scale ◽  
Domain Adaptation ◽  
Rapid Development ◽  
Synthetic Data ◽  
Feature Space ◽  
Imaging Informatics ◽  
Training Data ◽  
Domain Transformation

Introduction: There has been a rapid development of deep learning (DL) models for medical imaging. However, DL requires a large labeled dataset for training the models. Getting large-scale labeled data remains a challenge, and multi-center datasets suffer from heterogeneity due to patient diversity and varying imaging protocols. Domain adaptation (DA) has been developed to transfer the knowledge from a labeled data domain to a related but unlabeled domain in either image space or feature space. DA is a type of transfer learning (TL) that can improve the performance of models when applied to multiple different datasets. Objective: In this survey, we review the state-of-the-art DL-based DA methods for medical imaging. We aim to summarize recent advances, highlighting the motivation, challenges, and opportunities, and to discuss promising directions for future work in DA for medical imaging. Methods: We surveyed peer-reviewed publications from leading biomedical journals and conferences between 2017-2020, that reported the use of DA in medical imaging applications, grouping them by methodology, image modality, and learning scenarios. Results: We mainly focused on pathology and radiology as application areas. Among various DA approaches, we discussed domain transformation (DT) and latent feature-space transformation (LFST). We highlighted the role of unsupervised DA in image segmentation and described opportunities for future development. Conclusion: DA has emerged as a promising solution to deal with the lack of annotated training data. Using adversarial techniques, unsupervised DA has achieved good performance, especially for segmentation tasks. Opportunities include domain transferability, multi-modal DA, and applications that benefit from synthetic data.

scholarly journals A Review of Deep-Learning-Based Medical Image Segmentation Methods

2021 ◽  
Vol 13 (3) ◽  
pp. 1224
Author(s):  
Xiangbin Liu ◽  
Liping Song ◽  
Shuai Liu ◽  
Yudong Zhang
Keyword(s):  
Image Processing ◽  
Image Segmentation ◽  
Deep Learning ◽  
Medical Image ◽  
Rapid Development ◽  
Medical Image Segmentation ◽  
Deep Convolutional Neural Networks ◽  
Biomedical Image Processing ◽  
Data Set ◽  
Segmentation Methods

As an emerging biomedical image processing technology, medical image segmentation has made great contributions to sustainable medical care. Now it has become an important research direction in the field of computer vision. With the rapid development of deep learning, medical image processing based on deep convolutional neural networks has become a research hotspot. This paper focuses on the research of medical image segmentation based on deep learning. First, the basic ideas and characteristics of medical image segmentation based on deep learning are introduced. By explaining its research status and summarizing the three main methods of medical image segmentation and their own limitations, the future development direction is expanded. Based on the discussion of different pathological tissues and organs, the specificity between them and their classic segmentation algorithms are summarized. Despite the great achievements of medical image segmentation in recent years, medical image segmentation based on deep learning has still encountered difficulties in research. For example, the segmentation accuracy is not high, the number of medical images in the data set is small and the resolution is low. The inaccurate segmentation results are unable to meet the actual clinical requirements. Aiming at the above problems, a comprehensive review of current medical image segmentation methods based on deep learning is provided to help researchers solve existing problems.

scholarly journals Cloud Detection for Satellite Imagery Using Attention-Based U-Net Convolutional Neural Network

Symmetry ◽  
2020 ◽  
Vol 12 (6) ◽  
pp. 1056
Author(s):  
Yanan Guo ◽  
Xiaoqun Cao ◽  
Bainian Liu ◽  
Mei Gao
Keyword(s):  
Image Processing ◽  
Deep Learning ◽  
Network Architecture ◽  
Rapid Development ◽  
Satellite Image ◽  
Remote Sensing Data ◽  
Attention Mechanism ◽  
Detection Methods ◽  
Great Success ◽  
Cloud Detection

Cloud detection is an important and difficult task in the pre-processing of satellite remote sensing data. The results of traditional cloud detection methods are often unsatisfactory in complex environments or the presence of various noise disturbances. With the rapid development of artificial intelligence technology, deep learning methods have achieved great success in many fields such as image processing, speech recognition, autonomous driving, etc. This study proposes a deep learning model suitable for cloud detection, Cloud-AttU, which is based on a U-Net network and incorporates an attention mechanism. The Cloud-AttU model adopts the symmetric Encoder-Decoder structure, which achieves the fusion of high-level features and low-level features through the skip-connection operation, making the output results contain richer multi-scale information. This symmetrical network structure is concise and stable, significantly enhancing the effect of image segmentation. Based on the characteristics of cloud detection, the model is improved by introducing an attention mechanism that allows model to learn more effective features and distinguish between cloud and non-cloud pixels more accurately. The experimental results show that the method proposed in this paper has a significant accuracy advantage over the traditional cloud detection method. The proposed method is also able to achieve great results in the presence of snow/ice disturbance and other bright non-cloud objects, with strong resistance to disturbance. The Cloud-AttU model proposed in this study has achieved excellent results in the cloud detection tasks, indicating that this symmetric network architecture has great potential for application in satellite image processing and deserves further research.

scholarly journals Full 3D Microwave Breast Imaging Using a Deep-Learning Technique

2020 ◽  
Vol 6 (8) ◽  
pp. 80 ◽  
Author(s):  
Vahab Khoshdel ◽  
Mohammad Asefi ◽  
Ahmed Ashraf ◽  
Joe LoVetri
Keyword(s):  
Deep Learning ◽  
Breast Imaging ◽  
Synthetic Data ◽  
Roc Curves ◽  
Source Inversion ◽  
Operating Characteristics ◽  
3D Images ◽  
Size Number ◽  
Contrast Source Inversion ◽  
Learning Technique

A deep learning technique to enhance 3D images of the complex-valued permittivity of the breast obtained via microwave imaging is investigated. The developed technique is an extension of one created to enhance 2D images. We employ a 3D Convolutional Neural Network, based on the U-Net architecture, that takes in 3D images obtained using the Contrast-Source Inversion (CSI) method and attempts to produce the true 3D image of the permittivity. The training set consists of 3D CSI images, along with the true numerical phantom images from which the microwave scattered field utilized to create the CSI reconstructions was synthetically generated. Each numerical phantom varies with respect to the size, number, and location of tumors within the fibroglandular region. The reconstructed permittivity images produced by the proposed 3D U-Net show that the network is not only able to remove the artifacts that are typical of CSI reconstructions, but it also enhances the detectability of the tumors. We test the trained U-Net with 3D images obtained from experimentally collected microwave data as well as with images obtained synthetically. Significantly, the results illustrate that although the network was trained using only images obtained from synthetic data, it performed well with images obtained from both synthetic and experimental data. Quantitative evaluations are reported using Receiver Operating Characteristics (ROC) curves for the tumor detectability and RMS error for the enhancement of the reconstructions.

Classification and Separation of Diffraction Energy on Pre-Migration Seismic Data using Deep Learning

2020 ◽  
Author(s):  
Brydon Lowney ◽  
Ivan Lokmer ◽  
Gareth Shane O'Brien ◽  
Christopher Bean
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Deep Neural Network ◽  
Computational Cost ◽  
Synthetic Data ◽  
Real Data ◽  
Training Data ◽  
Separation Technique ◽  
Generative Adversarial Network ◽  
Analytical Separation

<p>Diffractions are a useful aspect of the seismic wavefield and are often underutilised. By separating the diffractions from the rest of the wavefield they can be used for various applications such as velocity analysis, structural imaging, and wavefront tomography. However, separating the diffractions is a challenging task due to the comparatively low amplitudes of diffractions as well as the overlap between reflection and diffraction energy. Whilst there are existing analytical methods for separation, these act to remove reflections, leaving a volume which contains diffractions and noise. On top of this, analytical separation techniques can be costly computationally as well as requiring manual parameterisation. To alleviate these issues, a deep neural network has been trained to automatically identify and separate diffractions from reflections and noise on pre-migration data.</p><p>Here, a Generative Adversarial Network (GAN) has been trained for the automated separation. This is a type of deep neural network architecture which contains two neural networks which compete against one another. One neural network acts as a generator, creating new data which appears visually similar to the real data, while a second neural network acts as a discriminator, trying to identify whether the given data is real or fake. As the generator improves, so too does the discriminator, giving a deeper understanding of the data. To avoid overfitting to a specific dataset as well as to improve the cross-data applicability of the network, data from several different seismic datasets from geologically distinct locations has been used in training. When comparing a network trained on a single dataset compared to one trained on several datasets, it is seen that providing additional data improves the separation on both the original and new datasets.</p><p>The automatic separation technique is then compared with a conventional, analytical, separation technique; plane-wave destruction (PWD). The computational cost of the GAN separation is vastly superior to that of PWD, performing a separation in minutes on a 3-D dataset in comparison to hours. Although in some complex areas the GAN separation is of a higher quality than the PWD separation, as it does not rely on the dip, there are also areas where the PWD outperforms the GAN separation. The GAN may be enhanced by adding more training data as well as by improving the initial separation used to create the training data, which is based around PWD and thus is imperfect and can introduce bias into the network. A potential for this is training the GAN entirely using synthetic data, which allows for a perfect separation as the points are known, however, it must be of sufficient volume for training and sufficient quality for real data applicability.</p>

scholarly journals Quantum Computing Aided Machine Learning Through Quantum State Fidelity

2021 ◽  
Author(s):  
Samuel A. Stein
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Quantum Computing ◽  
Deep Learning ◽  
Network Architecture ◽  
Rapid Development ◽  
Original Data ◽  
Quantum Circuit ◽  
Data Generation ◽  
Quantum Neural Network

Tremendous progress has been witnessed in artificial intelligence within the domain of neural network backed deep learning systems and its applications. As we approach the post Moore’s Law era, the limit of semiconductor fabrication technology along with a rapid increase in data generation rates have lead to an impending growing challenge of tackling newer and more modern machine learning problems. In parallel, quantum computing has exhibited rapid development in recent years. Due to the potential of a quantum speedup, quantum based learning applications have become an area of significant interest, in hopes that we can leverage quantum systems to solve classical problems. In this work, we propose a quantum deep learning architecture; we demonstrate our quantum neural network architecture on tasks ranging from binary and multi-class classification to generative modelling. Powered by a modified quantum differentiation function along with a hybrid quantum-classic design, our architecture encodes the data with a reduced number of qubits and generates a quantum circuit, loading it onto a quantum platform where the model learns the optimal states iteratively. We conduct intensive experiments on both the local computing environment and IBM-Q quantum platform. The evaluation results demonstrate that our architecture is able to outperform Tensorflow-Quantum by up to 12.51% and 11.71% for a comparable classic deep neural network on the task of classification trained with the same network settings. Furthermore, our GAN architecture runs the discriminator and the generator purely on quantum hardware and utilizes the swap test on qubits to calculate the values of loss functions. In comparing our quantum GAN, we note our architecture is able to achieve similar performance with 98.5% reduction on the parameter set when compared to classical GANs. With the same number of parameters, additionally, QuGAN outperforms other quantum based GANs in the literature for up to 125.0% in terms of similarity between generated distributions and original data sets.

scholarly journals A Deep Learning Architecture For 3D Mapping Urban Landscapes

2021 ◽  
Vol 11 (23) ◽  
pp. 11551
Author(s):  
Armando Levid Rodríguez-Santiago ◽  
José Aníbal Arias-Aguilar ◽  
Hiroshi Takemura ◽  
Alberto Elías Petrilli-Barceló
Keyword(s):  
Deep Learning ◽  
Network Architecture ◽  
Point Cloud ◽  
Three Dimensional ◽  
Aerial Image ◽  
Urban Landscapes ◽  
Feature Maps ◽  
Generative Adversarial Network ◽  
Outdoor Environments ◽  
2D Images

In this paper, an approach through a Deep Learning architecture for the three-dimensional reconstruction of outdoor environments in challenging terrain conditions is presented. The architecture proposed is configured as an Autoencoder. However, instead of the typical convolutional layers, some differences are proposed. The Encoder stage is set as a residual net with four residual blocks, which have been provided with the necessary knowledge to extract the feature maps from aerial images of outdoor environments. On the other hand, the Decoder stage is set as a Generative Adversarial Network (GAN) and called a GAN-Decoder. The proposed network architecture uses a sequence of the 2D aerial image as input. The Encoder stage works for the extraction of the vector of features that describe the input image, while the GAN-Decoder generates a point cloud based on the information obtained in the previous stage. By supplying a sequence of frames that a percentage of overlap between them, it is possible to determine the spatial location of each generated point. The experiments show that with this proposal it is possible to perform a 3D representation of an area flown over by a drone using the point cloud generated with a deep architecture that has a sequence of aerial 2D images as input. In comparison with other works, our proposed system is capable of performing three-dimensional reconstructions in challenging urban landscapes. Compared with the results obtained using commercial software, our proposal was able to generate reconstructions in less processing time, with less overlapping percentage between 2D images and is invariant to the type of flight path.

Sign in / Sign up

Export Citation Format

Share Document