检索结果-南通市图书馆

Using Cross Entropy as a Performance Metric for Quantifying Uncertainty in DNN Image Classifiers: An Application to Classification of Lung Cancer on CT Images

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Biomedical Science and Engineering 2024年第1期17卷 1-12页

作者： Eri Matsuyama Masayuki Nishiki Noriyuki Takahashi haruyuki Watanabe Faculty of Informatics University of Fukuchiyama Kyoto Japan Graduate School of Radiological Sciences International University of Health and Welfare Tochigi Japan School of Health Sciences Fukushima Medical University Fukushima Japan School of Radiological Technology Gunma Prefectural College of Health Sciences Gunma Japan

Cross entropy is a measure in machine learning and deep learning that assesses the difference between predicted and actual probability distributions. In this study, we propose cross entropy as a performance evaluation metric for image classifier models and apply it to the CT image classification of lung cancer. A convolutional neural network is employed as the deep neural network (DNN) image classifier, with the residual network (ResNet) 50 chosen as the DNN archi-tecture. The image data used comprise a lung CT image set. Two classification models are built from datasets with varying amounts of data, and lung cancer is categorized into four classes using 10-fold cross-validation. Furthermore, we employ t-distributed stochastic neighbor embedding to visually explain the data distribution after classification. Experimental results demonstrate that cross en-tropy is a highly useful metric for evaluating the reliability of image classifier models. It is noted that for a more comprehensive evaluation of model perfor-mance, combining with other evaluation metrics is considered essential. .

关键词： Cross Entropy Performance Metrics DNN Image Classifiers Lung Cancer Prediction Uncertainty

Tracing sediment transport history using mineralogical fingerprinting in a river basin with dams utilizing sediment sluicing

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

International Journal of Sediment Research 2023年第3期38卷 469-480页

作者： Kenichi Ito Motohide Matsunaga Tomoya Itakiyo Hiroyuki Oishi Kei Nukazawa Mitsuteru irie Yoshihiro Suzuki Center for International Relations University of Miyazaki Department of Civil and Environmental Engineering Faculty of Engineering University of Miyazaki Oita Prefectural Government West Japan Engineering Consultants Inc.

Sediment causes a serious problem in relation to dam function. A cooperative sediment sluicing operation has been under way since 2017 to prevent sediment from accumulating in dams in the Mimi River,Miyazaki, Japan. To achieve a smooth and stable operation, it is very important to determine the sediment source and a sediment transport system to maintain the dam’s function. In the current study, the source and transport of sediment from the Mimi River basin have been analyzed with X-ray diffraction(XRD) to reveal the peaks of mineral species. The sediment samples were collected in the Mimi River basin from the Tsukabaru Dam to the sea in an area including 4 dams and 5 tributaries. In addition, the minerals in samples collected in 2014, before the start of the sediment sluicing operation, and from 2018 to 2020, after the start of the operation, were analyzed. An evaluation of the similarity of the sediment at each sampling point based on the X-ray diffraction peaks of mineral species showed that sediment distributed upstream was transported downstream in the year when the sediment sluicing was *** result indicates that the sediment sluicing operation at dams ensured the continuity of the sediment distribution in the Mimi River basin. In addition, an investigation done in 2020 showed that sediment particles very similar to those of the upstream tributaries were deposited downstream because of extensive flooding caused by a large typhoon. The management of sediment transport has the greatest importance in a river basin where a dam is to be constructed. It is possible to trace the history of the sediment distribution and movement resulting from the operation of sediment sluicing by using sediment mineral analysis.

关键词： Sediment sluicing Sediment source X-ray diffraction (XRD) Mineralogical analysis Cluster analysis

Explainable Analysis of Deep Learning Models for Coronavirus Disease (COVID-19) Classification with Chest X-Ray Images: Towards Practical Applications

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Open Journal of Medical Imaging 2022年第3期12卷 83-102页

作者： Eri Matsuyama haruyuki Watanabe Noriyuki Takahashi Faculty of Informatics University of Fukuchiyama Kyoto Japan School of Radiological Technology Gunma Prefectural College of Health Sciences Gunma Japan School of Hearth Sciences Fukushima Medical University Fukushima Japan

In recent years, with numerous developments of convolutional neural network (CNN) classification models for medical diagnosis, the issue of misrecognition/misclassification has become more and more important. Thus, research on misrecognition/misclassification has been progressing. This study focuses on the problem of misrecognition/misclassification of CNN classification models for coronavirus disease (COVID-19) using chest X-ray images. We construct two models for COVID-19 pneumonia classification by fine-tuning ResNet-50 architecture, i.e., a model retrained with full-sized original images and a model retrained with segmented images. The present study demonstrates the uncertainty (misrecognition/misclassification) of model performance caused by the discrepancy in the shapes of images at the phase of model construction and that of clinical applications. To achieve it, we apply three XAI methods to demonstrate and explain the uncertainty of classification results obtained from the two constructed models assuming for clinical applications. Experimental results indicate that the performance of classification models cannot be maintained when the type of constructed model and the geometric shape of input images are not matched, which may bring about misrecognition in clinical applications. We also notice that the effect of adversarial attack might be induced if the method of image segmentation is not performed properly. The results suggest that the best approach to obtaining a highly reliable prediction in the classification of COVID-19 pneumonia is to construct a model using full-sized original images as training data and use full-sized original images as the input when utilized in clinical applications.

关键词： Explainable AI CNN CXR Image COVID-19 Pneumonia

A Breast Density Classification System for Mammography Considering Reliability Issues in Deep Learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Open Journal of Medical Imaging 2023年第3期13卷 63-83页

作者： Eri Matsuyama Megumi Takehara Noriyuki Takahashi haruyuki Watanabe Faculty of Informatics University of Fukuchiyama Kyoto Japan Breast Center Dokkyo Medical University Hospital Tochigi Japan School of Hearth Sciences Fukushima Medical University Fukushima Japan School of Radiological Technology Gunma Prefectural College of Health Sciences Gunma Japan

In a convolutional neural network (CNN) classification model for diagnosing medical images, transparency and interpretability of the model’s behavior are crucial in addition to high classification accuracy, and it is highly important to explicitly demonstrate them. In this study, we constructed an interpretable CNN-based model for breast density classification using spectral information from mammograms. We evaluated whether the model’s prediction scores provided reliable probability values using a reliability diagram and visualized the basis for the final prediction. In constructing the classification model, we modified ResNet50 and introduced algorithms for extracting and inputting image spectra, visualizing network behavior, and quantifying prediction ambiguity. From the experimental results, our proposed model demonstrated not only high classification accuracy but also higher reliability and interpretability compared to the conventional CNN models that use pixel information from images. Furthermore, our proposed model can detect misclassified data and indicate explicit basis for prediction. The results demonstrated the effectiveness and usefulness of our proposed model from the perspective of credibility and transparency.

关键词： Explainable AI t-SNE Entropy Wavelet Transform Mammogram

Advanced diffusion-weighted magnetic resonance imaging in the evaluation of white matter axons in patients with idiopathic normal pressure hydrocephalus

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Neural Regeneration Research 2017年第12期12卷 1974-1975页

作者： Masaaki Hori Kouhei Kamiya Ryusuke irie Department of Radiology Juntendo University School of MedicineTokyo Japan Department of Radiology The University of Tokyo Tokyo Japan

With the increasing application of regenerative medicine in vivo,non-invasive and accurate methods for estimating the white matter axons in neuronal tissue have become increasingly *** a non-invasive method for patients,magnetic resonance（MR）imaging has demonstrated potential as a promising tool for evaluating axons in *** particular,diffusion-weighted MR imaging（d MRI）and its applications,

关键词： Advanced diffusion-weighted magnetic white matter pressure hydrocephalus

Application of Rainbow Schlieren Deflectometry for Axisymmetric Supersonic Jets (Comparison of Experiments with Numerical Analysis)

在线全文

学校读者我要写书评

暂无评论

Journal of Thermal Science 2010年第3期19卷 218-221页

作者： Hideki Yamamoto Masayuki irie Yoshiaki Miyazato Kazuyasu Matsuo Department of Mechanical Systems Engineering Faculty of Environmental EngineeringThe University of Kitakyushu1-1 HibikinoWakamatsu-kuKitakyushuFukuoka 808-0135Japan

This paper analyzes the correctly-expanded supersonic jet from a convergent-divergent axisymmetric nozzle by using numerical simulation of turbulent *** the calculated density distributions in this flow are compared with the present experimental data using rainbow schlieren *** value of the density from the experimental data agrees well with the results calculated by this ***,the present method of the measurement using rainbow schlieren deflectometry is useful for the measurement of the density of the correctly-expanded supersonic jet.

关键词：超音速射流实验数据轴对称数值分析彩虹偏折纹影数值模拟计算

A Purely Elastic Instability and Mixing Enhancement in a 3D Curvilinear Channel Flow

在线全文

学校读者我要写书评

暂无评论

Chinese Physics Letters 2012年第9期29卷 105-109页

作者：李凤臣张红娜曹阳 KUNUGI Tomoaki KINOSHITA haruyuki OSHIMA Marie School of Energy Science and Engineering Harbin Institute of TechnologyHarbin 150001 Department of Nuclear Engineering Kyoto UniversityKyoto 606-8501Japan Institute of Industrial Science The University of TokyoTokyo 153-8505Japan

We report the detailed information of a purely elastic instability and its mixing enhancement effect in a viscoelastic fluid flow driven by a constant pressure gradient along a curvilinear channel obtained for the first time from the three-dimensional direct numerical simulations(DNSs)of this kind of flow ***-dimensional unstable flow structures similar to those visualized in experiments are realized through the *** inception of elastic instability,conformation of microstructures,statistical and dynamic information of velocity field and the mixing enhancement effect caused by elastic instability are then discussed.

关键词： mixing flow Elastic

Characteristics of Spherical Shock Wave and Circular Pulse Jet Generated by Discharge of Propagating Shock Wave at Open End of Tube

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Thermal Science 2003年第3期12卷 245-249页

作者： Tsukasa irie Tsuyoshi Yasunobu Hideo Kashimura Toshiaki Setoguchi Kazuyasu Matsuo Department of Mechanical Engineering Kitakyushu National College of Technology ShiiKokuraminami-ku Kitakyushu 802-0985 Japan Department of Control and Information Systems Engineering Kitakyushu National College of Technology Shii Kokuraminami-ku Kitakyushu 802-0985 Japan Department of Mechanical Engineering Saga University Honjo Saga 840-8502 Japan Department of Mechanical Systems and Environmental Engineering Univ. of Kitakyushu1-1 Hibikino Wakamatsu-ku Kitakyushu808-0135 Japan

When the shock wave propagating in the straight circular tube reaches at the open end, the impulsive wave is generated by the emission of a shock wave from an open end, and unsteady pulse jet is formed near the open end behind the impulsive wave under the specific condition. The pulse jet transits to spherical shock wave with the increase in the strength of shock wave. The strength is dependent on the Mach number of shock wave, which attenuates by propagation distance from the open end. In this study, the mechanism of generating the unsteady pulse jet, the characteristics of the pressure distribution in the flow field and the emission of shock wave from straight circular tube which has the infinite flange at open end are analyzed numerically by the TVD method. Strength of spherical shock wave, relation of shock wave Mach number,distance decay of spherical shock wave and directional characteristics are clarified.

关键词： shock wave impulsive wave numerical analysis spherical shock wave TVD method

Characteristics of the Mach Disk in the Underexpanded Jet in which the Back Pressure Continuously Changes with Time

在线全文

学校读者我要写书评

暂无评论

Journal of Thermal Science 2003年第2期12卷 132-137页

作者： T.irie T.Yasunobu H.Kashimura T.Setoguchi Department of Mechanical Engineering Kitakyushu National College of TechnologyShiiKokuraminami-kuKitakyushu 802-0985JapanDepartment of Control and Information Systems EngineeringKitakyushu National College of TechnologyShiiKokuraminami-kuKitakyushu 802-0985JapanDepartment of Control and Information Systems EngineeringKitakyushu National College of TechnologyShiiKokuraminami-kuKitakyushu 802-0985JapanDepartment of Mechanical EngineeringSaga UniversityHonjoSaga 840-8502Japan

When the high-pressure gas is exhausted to the vacuum chamber from the nozzle,the underexpanded supersonic jet contained with the Mach disk is generally *** eventual purpose of this study is to clarify the unsteady phenomenon of the underexpanded free jet when the back pressure continuously changes with *** characteristic of the Mach disk has been clarified in consideration of the diameter and position of it by the numerical analysis in this *** sonic jet of the exit Mach number Me=1 is assumed and the axisymmetric conservational equation is solved by the TVD method in the numerical *** diameter and position of the Mach disk differs with the results of a steady jet and the influence on the continuously changing of the back pressure is evidenced from the comparison with the case of steady supersonic jet.

关键词：不完全膨胀喷流音速喷流规则反射马赫反射马赫圆盘 TVD法冲击波