• Korean Journal of Radiology
• /
• v.22 no.3
• /
• pp.476-488
• /
• 2021

Objective: We aimed to develop a deep neural network for segmenting lung parenchyma with extensive pathological conditions on non-contrast chest computed tomography (CT) images. Materials and Methods: Thin-section non-contrast chest CT images from 203 patients (115 males, 88 females; age range, 31-89 years) between January 2017 and May 2017 were included in the study, of which 150 cases had extensive lung parenchymal disease involving more than 40% of the parenchymal area. Parenchymal diseases included interstitial lung disease (ILD), emphysema, nontuberculous mycobacterial lung disease, tuberculous destroyed lung, pneumonia, lung cancer, and other diseases. Five experienced radiologists manually drew the margin of the lungs, slice by slice, on CT images. The dataset used to develop the network consisted of 157 cases for training, 20 cases for development, and 26 cases for internal validation. Two-dimensional (2D) U-Net and three-dimensional (3D) U-Net models were used for the task. The network was trained to segment the lung parenchyma as a whole and segment the right and left lung separately. The University Hospitals of Geneva ILD dataset, which contained high-resolution CT images of ILD, was used for external validation. Results: The Dice similarity coefficients for internal validation were 99.6 ± 0.3% (2D U-Net whole lung model), 99.5 ± 0.3% (2D U-Net separate lung model), 99.4 ± 0.5% (3D U-Net whole lung model), and 99.4 ± 0.5% (3D U-Net separate lung model). The Dice similarity coefficients for the external validation dataset were 98.4 ± 1.0% (2D U-Net whole lung model) and 98.4 ± 1.0% (2D U-Net separate lung model). In 31 cases, where the extent of ILD was larger than 75% of the lung parenchymal area, the Dice similarity coefficients were 97.9 ± 1.3% (2D U-Net whole lung model) and 98.0 ± 1.2% (2D U-Net separate lung model). Conclusion: The deep neural network achieved excellent performance in automatically delineating the boundaries of lung parenchyma with extensive pathological conditions on non-contrast chest CT images.

• Journal of the Optical Society of Korea
• /
• v.11 no.3
• /
• pp.101-107
• /
• 2007

In this paper, we have presented the results of Wavelength Division Multiplexing-Passive Optical Network (WDM-PON) based fiber-to-the-home (FTTH) field trial test which was held in the city of Gwangju. We have implemented an injection locked Fabry-Perot Laser Diode (FP-LD) based WDM-PON system and reliably delivered Internet Protocol TV (IP-TV), networked Personal Video Recorder (N-PVR), High-Definition Video on Demand (HD-VoD), Education on Demand (EoD) and Internet service as FTTH service through the system during the field trial test. We have also verified that the WDM-PON system worked well to provide quality of service (QoS) guaranteed 100Mbps bandwidth per subscriber. Furthermore, we have presented network designing issues in Outside Plant (OSP) and Customer Premises Network (CPN) that should be overcome to efficiently deploy FTTH service. Finally, based on the field trial test results, we proposed FTTH service deployment strategies.

• Structural Engineering and Mechanics
• /
• v.49 no.6
• /
• pp.687-703
• /
• 2014

This paper develops a new nonlinear model for active control of three-dimensional (3D) irregular building structures. Both geometrical and material nonlinearities with a neuro-controller training algorithm are applied to a multi-degree-of-freedom 3D system. Two dynamic assembling motions are considered simultaneously in the control model such as coupling between torsional and lateral responses of the structure and interaction between the structural system and the actuators. The proposed control system and training algorithm of the structural system are evaluated by simulating the responses of the structure under the El-Centro 1940 earthquake excitation. In the numerical example, the 3D three-story structure with linear and nonlinear stiffness is controlled by a trained neural network. The actuator dynamics, control time delay and incident angle of earthquake are also considered in the simulation. Results show that the proposed control algorithm for 3D buildings is effective in structural control.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2002.03a
• /
• pp.759-766
• /
• 2002

In the paper the authors describe the development of ITIS(Intelligent Tunneling Information System) for the Purpose of applying the 3D visualization technique, GIS, AI(Artificial Intelligence) to tunnel design and construction. VR(Virtual Reality) and 3D visualization techniques are applied in order to develope the 3D model of characteristics and structures of ground and rock mass. Database for all the materials related to site investigation and tunnel construction is developed using GIS technique. AI technique such as fuzzy theory and neural network is applied to predict ground settlement, decide tunnel support method and estimate ground and rock mass properties according to tunnel excavation steps. ITIS can help to inform various necessary tunnel information to engineers quickly and manage tunnel using acquired information based on D/B.

• Proceedings of the KIEE Conference
• /
• 2001.10a
• /
• pp.131-133
• /
• 2001

2D analysis is impossible for Hybrid stepping motor(HSM) and 3D analysis is necessary because the permanent magnet is magnetized to the axial direction. In this paper, the characteristics of HSM are analyzed by using 3D equivalent magnetic circuit network method(3D EMCNM). In addition, the trapezoidal element is introduced for the exact permeance calculation in the complex shape machinery such as HSM. The magnetic saturation of core is considered.

• Convergence Security Journal
• /
• v.16 no.5
• /
• pp.41-48
• /
• 2016

In this paper we propose a new method of security visualization, VisFlow, using traffic flows to solve the problems of existing traffic flows based visualization techniques that were a loss of end-to-end semantics of communication, reflection problem by symmetrical address coordinates space, and intuitive loss problem in mass of traffic. VisFlow, a simple and effective security visualization interface, can do a real-time analysis and monitoring the situation in the managed network with visualizing a variety of network behavior not seen in the individual traffic data that can be shaped into patterns. This is a way to increase the intuitiveness and usability by identifying the role of nodes and by visualizing the highlighted or simplified information based on their importance in 2D/3D space. In addition, it monitor the network security situation as a way to increase the informational effectively using the asymmetrical connecting line based on IP addresses between pairs of nodes. Administrator can do a real-time analysis and monitoring the situation in the managed network using VisFlow, it makes to effectively investigate the massive traffic data and is easy to intuitively understand the entire network situation.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.16 no.2 s.93
• /
• pp.189-193
• /
• 2005

In this paper, we propose a direct conversion double-balanced down-converter fer MB-OFDM W system, which is implemented using $0.18\;{\mu}m$ CMOS technology and its measurement results are shown. The proposed down-converter adopts a resistive current-source instead of general transconductance stage using MOS transistor to achieve wideband characteristics over RF input frequency band $3\~5\;GHz$ with good gain flatness. The measured conversion gain is more than +3 dB, and gain flatness is less than 3 dB for three UWB channels. The dc consumption of this work is only 0.89 mA from 1.8 V power supply, leading to the low-power W application.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.4
• /
• pp.234-240
• /
• 2022

This paper proposes a customized AI exercise recommendation service for balancing the relative amount of exercise according to the working environment by each occupation. WISDM database is collected by using acceleration and gyro sensors, and is a dataset that classifies physical activities into 18 categories. Our system recommends a adaptive exercise using the analyzed activity type after classifying 18 physical activities into 3 physical activities types such as whole body, upper body and lower body. 1 Dimensional convolutional neural network is used for classifying a physical activity in this paper. Proposed model is composed of a convolution blocks in which 1D convolution layers with a various sized kernel are connected in parallel. Convolution blocks can extract a detailed local features of input pattern effectively that can be extracted from deep neural network models, as applying multi 1D convolution layers to input pattern. To evaluate performance of the proposed neural network model, as a result of comparing the previous recurrent neural network, our method showed a remarkable 98.4% accuracy.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.48 no.2
• /
• pp.1-5
• /
• 2011

For measuring quad-band module system using WiBro network, frequency converter was developed. The size of the fabricated frequency converter is $3.1cm{\times}3.1cm{\times}0.4cm$. Noise figure of the receiver part of the frequency converter was 2.62 ~ 3.45 dB, EVM of that is -37.5 dB ~ -34.5 dB. And EVM of the transmission part was -42.5 ~ -35.5 dB. Quad-band module was fabricated with the developed frequency converter. Testing the quad-band module in 2.3 GHz WiBro network results the excellent internet connection for 2.5 GHz, 3.5 GHz and 5.5 GHz band.

• Journal of Information Processing Systems
• /
• v.14 no.1
• /
• pp.176-190
• /
• 2018

This paper presents a novel convolutional neural network based multi-feature fusion learning method for non-rigid 3D model retrieval, which can investigate the useful discriminative information of the heat kernel signature (HKS) descriptor and the wave kernel signature (WKS) descriptor. At first, we compute the 2D shape distributions of the two kinds of descriptors to represent the 3D model and use them as the input to the networks. Then we construct two convolutional neural networks for the HKS distribution and the WKS distribution separately, and use the multi-feature fusion layer to connect them. The fusion layer not only can exploit more discriminative characteristics of the two descriptors, but also can complement the correlated information between the two kinds of descriptors. Furthermore, to further improve the performance of the description ability, the cross-connected layer is built to combine the low-level features with high-level features. Extensive experiments have validated the effectiveness of the designed multi-feature fusion learning method.