检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Huang, Ziyi Zhang, Haofeng Laine, Andrew Angelini, Elsa Hendon, Christine Gan, Yu Department of Electrical Engineering Columbia University New YorkNY United States Department of Industrial Engineering and Operations Research Columbia University New YorkNY United States Department of Biomedical Engineering Columbia University New YorkNY United States NIHR Imperial Biomedical Research Centre ITMAT Data Science Group Imperial College London United Kingdom Department of Electrical and Computer Engineering University of Alabama TuscaloosaAL United States

Supervised deep learning performance is heavily tied to the availability of high-quality labels for training. Neural networks can gradually overfit corrupted labels if directly trained on noisy datasets, leading to severe performance degradation at test time. In this paper, we propose a novel deep learning framework, namely Co-Seg, to collaboratively train segmentation networks on datasets which include low-quality noisy labels. Our approach first trains two networks simultaneously to sift through all samples and obtain a subset with reliable labels. Then, an efficient yet easily-implemented label correction strategy is applied to enrich the reliable subset. Finally, using the updated dataset, we retrain the segmentation network to finalize its parameters. Experiments in two noisy labels scenarios demonstrate that our proposed model can achieve results comparable to those obtained from supervised learning trained on the noise-free labels. In addition, our framework can be easily implemented in any segmentation algorithm to increase its robustness to noisy labels. Copyright © 2021, The Authors. All rights reserved.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

When Neural Networks Fail to Generalize? A Model Sensitivity Perspective

arXiv

引用

arXiv 2022年

作者： Zhang, Jiajin Chao, Hanqing Dhurandhar, Amit Chen, Pin-Yu Tajer, Ali Xu, Yangyang Yan, Pingkun Department of Biomedical Engineering Center for Biotechnology and Interdisciplinary Studies Rensselaer Polytechnic Institute TroyNY United States IBM Thomas J. Watson Research Center Yorktown HeightsNY United States Department of Electrical Computer and Systems Engineering Rensselaer Polytechnic Institute TroyNY United States Department of Mathematical Sciences Rensselaer Polytechnic Institute TroyNY United States

Domain generalization (DG) aims to train a model to perform well in unseen domains under different distributions. This paper considers a more realistic yet more challenging scenario, namely Single Domain Generalization (Single-DG), where only a single source domain is available for training. To tackle this challenge, we first try to understand when neural networks fail to generalize? We empirically ascertain a property of a model that correlates strongly with its generalization that we coin as "model sensitivity". Based on our analysis, we propose a novel strategy of Spectral Adversarial Data Augmentation (SADA) to generate augmented images targeted at the highly sensitive frequencies. Models trained with these hard-to-learn samples can effectively suppress the sensitivity in the frequency space, which leads to improved generalization performance. Extensive experiments on multiple public datasets demonstrate the superiority of our approach, which surpasses the state-of-the-art single-DG methods by up to 2.55%. The source code is available at https://***/DIAL-RPI/Spectral-Adversarial-Data-Augmentation. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Kinetics of the Luminescence of Isoelectronic Rare-Earth Ions In III-V Semiconductors

引用

MRS Online Proceedings Library 2020年第1期301卷 397-402页

作者： H. J. Lozykowski Department of Electrical and Computer Engineering and Condensed Matter & Surface Sciences Program Ohio University Athens USA

In this work we have developed a model for the kinetics of the energy transfer from the host lattice to the localized core excited states of rare earth isoelectronic structured traps (REI-trap). We have derive a set of differential equations for semi-insulating semiconductor governing the kinetics of rare earth luminescence. The numerically simulated rise and decay times of luminescence show a good quantitative agreement with the experimental data obtained for InP:Yb, over a wide range of generation rates.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Recognition of food-texture attributes using an in-ear microphone

arXiv

引用

arXiv 2021年

作者： Mars, Monica Diou, Christos Papapanagiotou, Vasileios van den Boer, Janet Delopoulos, Anastasios Multimedia Understanding Group Dpt. Electrical and Computer Engineering Faculty of Engineering Aristotle University of Thessaloniki Greece Department of Informatics and Telematics Harokopio University of Athens Greece Dpt. of Biomedical Signals and Systems Faculty of Electrical Engineering Mathematics and Computer Science University of Twente Netherlands Division of Human Nutrition and Health Wageningen University Netherlands

Food texture is a complex property;various sensory attributes such as perceived crispiness and wetness have been identified as ways to quantify it. Objective and automatic recognition of these attributes has applications in multiple fields, including health sciences and food engineering. In this work we use an in-ear microphone, commonly used for chewing detection, and propose algorithms for recognizing three food-texture attributes, specifically crispiness, wetness (moisture), and chewiness. We use binary SVMs, one for each attribute, and propose two algorithms: one that recognizes each texture attribute at the chew level and one at the chewing-bout level. We evaluate the proposed algorithms using leave-one-subject-out cross-validation on a dataset with 9 subjects. We also evaluate them using leave-one-food-type-out cross-validation, in order to examine the generalization of our approach to new, unknown food types. Our approach performs very well in recognizing crispiness (0.95 weighted accuracy on new subjects and 0.93 on new food types) and demonstrates promising results for objective and automatic recognition of wetness and chewiness. Copyright © 2021, The Authors. All rights reserved.

关键词： Wearable technology

来源：评论

学校读者我要写书评

暂无评论

Spark Plasma Sintering of Complex Metal and Ceramic Structures Produced by Robocasting

SSRN

引用

SSRN 2022年

作者： Brucculeri, Riccardo Airoldi, Lorenzo Baldini, Primo Vigani, Barbara Rossi, Silvia Morganti, Simone Auricchio, Ferdinando Anselmi-Tamburini, Umberto Department of Chemistry University of Pavia V.le Taramelli 12 Pavia 27100 Italy Department of Drug Sciences University of Pavia V.le Taramelli 12 Pavia 27100 Italy Department of Civil Engineering and Architecture University of Pavia Via Ferrata 3 Pavia 27100 Italy Department of Electrical Computer and Biomedical Engineering University of Pavia Via Ferrata 5 Pavia 27100 Italy

The use of approaches alternative to laser fusion for the additive manufacturing of metals has often been hampered by the long sintering cycles these techniques require. Typical sintering cycles require heating at temperatures above 80% of the melting point for several hours or even tens of hours. The process is time and energy-consuming, particularly when high-melting materials are involved. Applying pressure can drastically reduce the time and temperature required for densification. Recently, a particular kind of pressure-assisted sintering process known as Spark Plasma Sintering (SPS) or Field Assisted Sintering (FAST) received considerable attention in academia and industry due to its ability to drastically reduce the time and temperatures required for the densification. However, conventional SPS/FAST techniques use uniaxial pressure and cannot be directly applied to the densification of objects presenting a complex geometry. This work shows how a modified SPS/FAST setup, operating in pseudo-isostatic mode, can be used for debinding and sinter objects produced by robocasting. This approach can be applied to metals and many materials, including metal-based and ceramic-based composites, if their geometry does not include closed cavities. Depending on the characteristics of the pressure-transfer medium, some level of anisotropy in the volume reduction associated with the densification can be observed, but it can easily be corrected by modifying the starting geometry appropriately. The time required for the debinding and sintering of most metals and ceramic materials can be reduced to less than an hour, improving the rapid prototyping capabilities of robocasting considerably. This approach offers an alternative to the additive manufacturing of a wide range of inorganic materials, presenting a relatively low cost, high material flexibility, and low environmental impact. © 2022, The Authors. All rights reserved.

关键词： Spark plasma sintering

来源：评论

学校读者我要写书评

暂无评论

Mirrored Arrays for Direction-of-Arrival Estimation

Mirrored Arrays for Direction-of-Arrival Estimation

引用

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： Dong Zhu Gang Li Xiao-Ping Zhang Department of Electronic Engineering Tsinghua University China Department of Electrical Computer & Biomedical Engineering Ryerson University Canada

ISBN: (数字)9781509066315

ISBN: (纸本)9781509066322

A mirrored array configuration, which consists of an ordinary linear array and a reflector, is proposed. The mirrored arrays can provide composable degrees of freedom (DOF) for array processing by combined measurements with the help of mechanical adjustment (e.g., movement or rotation). This composability naturally leads to DOF enhancement and facilitates its applications in direction-of-arrival (DOA) estimation. The feasibility and effectiveness of the proposed arrays are demonstrated through numerical analysis.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Microchip-based structure determination of low-molecular weight proteins using cryo-electron microscopy

引用

Nanoscale 2021年第15期13卷 7285-7293页

作者： Casasanta, Michael A. Jonaid, G.M. Kaylor, Liam Luqiu, William Y. Solares, Maria J. Schroen, Mariah L. Dearnaley, William J. Wilson, Jarad Dukes, Madeline J. Kelly, Deborah F. Department of Biomedical Engineering Pennsylvania State University University Park 16802 PA United States Materials Research Institute Pennsylvania State University University Park 16802 PA United States Bioinformatics and Genomics Graduate Program Huck Institutes of the Life Sciences Pennsylvania State University University Park 16802 PA United States Molecular Cellular and Integrative Biosciences Graduate Program Huck Institutes of the Life Sciences Pennsylvania State University University Park 16802 PA United States Department of Electrical and Computer Engineering Duke University Durham 27708 NC United States RayBiotech Life Peachtree Corners 30092 GA United States Applications Science Protochips Inc Morrisville 27560 NC United States

Interest in cryo-Electron Microscopy (EM) imaging has skyrocketed in recent years due to its pristine views of macromolecules and materials. As advances in instrumentation and computing algorithms spurred this progress, there is renewed focus to address specimen-related challenges. Here we contribute a microchip-based toolkit to perform complementary structural and biochemical analysis on low-molecular weight proteins. As a model system, we used the SARS-CoV-2 nucleocapsid (N) protein (48 kDa) due to its stability and important role in therapeutic development. Cryo-EM structures of the N protein monomer revealed a flexible N-terminal "top hat"motif and a helical-rich C-terminal domain. To complement our structural findings, we engineered microchip-based immunoprecipitation assays that led to the discovery of the first antibody binding site on the N protein. The data also facilitated molecular modeling of a variety of pandemic and common cold-related coronavirus proteins. Such insights may guide future pandemic-preparedness protocols through immuno-engineering strategies to mitigate viral outbreaks. This journal is © The Royal Society of Chemistry.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Morphology Focused Diffusion Probabilistic Model for Synthesis of Histopathology Images

arXiv

引用

arXiv 2022年

作者： Moghadam, Puria Azadi Van Dalen, Sanne Martin, Karina C. Lennerz, Jochen Yip, Stephen Farahani, Hossein Bashashati, Ali Department of Electrical and Computer Engineering University of British Columbia Canada Eindhoven University of Technology Netherlands Department of Pathology and Laboratory Medicine University of British Columbia Canada Center for Integrated Diagnostics Department of Pathology Massachusetts General Hospital United States Department of Pathology Harvard Medical School United States Molecular Oncology BC Cancer Agency Canada School of Biomedical Engineering University of British Columbia Canada

Visual microscopic study of diseased tissue by pathologists has been the cornerstone for cancer diagnosis and prognostication for more than a century. Recently, deep learning methods have made significant advances in the analysis and classification of tissue images. However, there has been limited work on the utility of such models in generating histopathology images. These synthetic images have several applications in pathology including utilities in education, proficiency testing, privacy, and data sharing. Recently, diffusion probabilistic models were introduced to generate high quality images. Here, for the first time, we investigate the potential use of such models along with prioritized morphology weighting and color normalization to synthesize high quality histopathology images of brain cancer. Our detailed results show that diffusion probabilistic models are capable of synthesizing a wide range of histopathology images and have superior performance compared to generative adversarial networks. Copyright © 2022, The Authors. All rights reserved.

关键词： Diffusion

来源：评论

学校读者我要写书评

暂无评论

GAN-Based Realistic Bone Ultrasound Image and Label Synthesis for Improved Segmentation 23rd

GAN-Based Realistic Bone Ultrasound Image and Label Synthesi...

引用

23rd International Conference on Medical Image Computing and computer-Assisted Intervention, MICCAI 2020

作者： Alsinan, Ahmed Z. Rule, Charles Vives, Michael Patel, Vishal M. Hacihaliloglu, Ilker Department of Electrical and Computer Engineering Rutgers University PiscatawayNJ United States Department of Computer Science Rutgers University PiscatawayNJ United States Department of Orthopedics Rutgers New Jersey Medical School NewarkNJ United States Department of Electrical and Computer Engineering Johns Hopkins University BaltimoreMD United States Department of Biomedical Engineering Rutgers University PiscatawayNJ United States Department of Radiology Rutgers University Robert Wood Johnson Medical School New BrunswickNJ United States

ISBN: (纸本)9783030597245

To provide a safe alternative, for intra-operative fluoroscopy, ultrasound (US) has been investigated as an alternative safe imaging modality for various computer assisted orthopedic surgery (CAOS) procedures. However, low signal to noise ratio, imaging artifacts and bone surfaces appearing several millimeters (mm) in thickness have hindered the wide spread application of US in CAOS. In order to provide a solution for these problems, research has focused on the development of accurate, robust and real-time bone segmentation methods. Most recently methods based on deep learning have shown very promising results. However, scarcity of bone US data introduces significant challenges when training deep learning models. In this work, we propose a computational method, based on a novel generative adversarial network (GAN) architecture, to (1) produce synthetic B-mode US images and (2) their corresponding segmented bone surface masks in real-time. We show how a duality concept can be implemented for such tasks. Armed by two convolutional blocks, referred to as self-projection and self-attention blocks, our proposed GAN model synthesizes realistic B-mode bone US image and segmented bone masks. Quantitative and qualitative evaluation studies are performed on 1235 scans collected from 27 subjects using two different US machines to show comparison results of our model against state-of-the-art GANs for the task of bone surface segmentation using U-net. © 2020, Springer Nature Switzerland AG.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Epi-illumination tomographic quantitative phase imaging identifies brain tumor margins

Epi-illumination tomographic quantitative phase imaging iden...

引用

Label-free biomedical Imaging and Sensing, LBIS 2021

作者： Costa, Paloma Casteleiro Guang, Zhe Ledwig, Patrick Zhang, Zhaobin Neill, Stewart Olson, Jeffrey J. Robles, Francisco E. School of Electrical and Computer Engineering Georgia Institute of Technology AtlantaGA30332 United States Wallace H. Coulter Department of Biomedical Engineering Georgia Institute of Technology Emory University AtlantaGA30332 United States Winship Cancer Institute Emory University AtlantaGA30322 United States Department of Neurosurgery Emory University School of Medicine AtlantaGA30322 United States Department of Pathology and Laboratory Medicine Emory University School of Medicine AtlantaGA30322 United States

ISBN: (纸本)9781510641457

The first-line treatment for brain cancer is surgery, which focuses on maximizing the percentage of the tumor removed during surgery (i.e., extent of resection) while minimizing damage to healthy brain tissue. Data show that extent of resection is one of the most critical factors associated with prolonged survival. However, differentiating between tumor and healthy tissue intraoperatively remains a significant clinical challenge, resulting in an exceedingly low 5-year survival rate of only ~35%. In this work, we show that quantitative oblique back illumination microscopy (qOBM), a novel label-free optical imaging technique that achieves tomographic quantitative phase imaging (QPI) in thick scattering samples, clearly differentiates between tumor and healthy tissue. Using a 9L gliosarcoma rat tumor model, we show that quantitative image features from qOBM provide a robust set of biomarkers for disease. In addition, tumor regions, including diffuse tumor, and healthy brain structures, show excellent structural agreement with H&E stained and sliced brightfield images, the gold standard for cancer detection. The unique attribute of qOBM—low-cost, easy-to-use, label-free, and real-time—make this technology ideally suited to help guide neurosurgery and address this important unmet need. Here we describe our free-space qOBM system and present quantitative results from the 9L gliosarcoma rat tumor model. © 2021 SPIE.

关键词： Brain

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：