• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.10
• /
• pp.2768-2787
• /
• 2023

As an important research direction of the application of computer science in the medical field, the automatic generation technology of radiology report has attracted wide attention in the academic community. Because the proportion of normal regions in radiology images is much larger than that of abnormal regions, words describing diseases are often masked by other words, resulting in significant feature loss during the calculation process, which affects the quality of generated reports. In addition, the huge difference between visual features and semantic features causes traditional multi-modal fusion method to fail to generate long narrative structures consisting of multiple sentences, which are required for medical reports. To address these challenges, we propose a multi-feature optimization Transformer (MFOT) for generating radiology reports. In detail, a multi-dimensional mapping attention (MDMA) module is designed to encode the visual grid features from different dimensions to reduce the loss of primary features in the encoding process; a feature pre-fusion (FP) module is constructed to enhance the interaction ability between multi-modal features, so as to generate a reasonably structured radiology report; a detail enhanced attention (DEA) module is proposed to enhance the extraction and utilization of key features and reduce the loss of key features. In conclusion, we evaluate the performance of our proposed model against prevailing mainstream models by utilizing widely-recognized radiology report datasets, namely IU X-Ray and MIMIC-CXR. The experimental outcomes demonstrate that our model achieves SOTA performance on both datasets, compared with the base model, the average improvement of six key indicators is 19.9% and 18.0% respectively. These findings substantiate the efficacy of our model in the domain of automated radiology report generation.

• Journal of The Korean Association of Information Education
• /
• v.16 no.4
• /
• pp.483-488
• /
• 2012

As propagation of mobile smart devices is widespread, it is an observable trend that the cases of utilizing them are increasing in the school programs, and it is also anticipated that they will be very important part of the educational equipment in near future. For this reason the department of education and science technology has announced a medium and long term project on the education with smart device, which is undergoing the preparation stage, and the various academic and industrial institutes have actively produced the related research results and the application prototypes. In this paper we propose a framework on design and implementation of a visual context recognition system for educational purpose usable in the school program by utilizing a module for recognition of the texts embedded in the image captured by video camera from mobile smart device. The system proposed in this paper is consisted of the four modules, such as, image acquisition, image processing, information extraction, and knowledge representation, which are explained in details with the practical examples.

• Journal of the Microelectronics and Packaging Society
• /
• v.9 no.2
• /
• pp.45-48
• /
• 2002

LTCC system has many kinds of advantages, like low loss, low cost of process, stability of process etc. But it is so hard to adjust the characteristics of passives in ceramic module after fabrication. So the exact prediction of behavior of components in high frequency region upper than GHz must be made. In this procedure, the exact measurement is need. In this study, many kinds of measurement Jigs are compared and optimized, and measurement methods of each parameter are designed.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2016.02a
• /
• pp.424.1-424.1
• /
• 2016

태양전지 모듈의 봉지재로 사용되는 EVA(Ethylene-Vinyl acetate)는 외부의 환경으로부터 태양전지를 보호해주는 역할을 한다. 하지만 frame의 실링재나 glass edge 부분의 결함으로 인해 수분이 침투되어 모듈의 내부자재에 영향을 미치게 되며, 모듈의 효율 감소나 수명 단축의 원인이 된다. 이러한 결함을 줄이기 위해서는 EVA의 crosslinking degree, 절연 저항 특성을 향상시키는 것이 중요하다. 이에 본 연구에서는 EVA의 crosslinking degree를 보기 위해 soxhlet extraction method를 이용하여 gel content를 측정하였으며, 라미네이션 과정 중 curing time에 따른 gel content 변화를 확인하였다. 또한 이 실험을 바탕으로 gel content에 따른 절연 저항을 측정하여 EVA의 crosslinking degree에 의한 절연 저항 특성을 확인하였다.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.57 no.1
• /
• pp.142-147
• /
• 2008

Abdominal circumference changes due to breathing by the respiratory muscle activity such as diaphragm, which would partially represent the lung volume variation. The present study introduced conductive rubber molded in a cord shape incorporated with a patient's pants. The conductive rubber cord operated as a displacement transducer to measure the lung or abdominal volume changes. Signal extraction circuitry was developed to obtain the volume and its derivative(or the flow) signals followed by wireless transmission based on the Zigbee communication protocol in a size of $65mm{\times}105mm$ easily put in pocket. Breathing frequency was accurately evaluated and breath pattern analysis seemed feasible, since respiratory behaviours such as maximal inspiration and cough were well identified. Remote wireless receiver module also enabled to monitor both volume and flow signals during resting breathing on a PC terminal.

• KIEAE Journal
• /
• v.10 no.4
• /
• pp.75-80
• /
• 2010

Generally, ground-source heat pump (GSHP) systems have a higher performance than conventional air-source systems. However, the major fault of GSHP systems is their expensive boring costs. Therefore, it is important issue that to reduce initial cost and ensure stability of system through accurate prediction of the heat extraction and injection rates of the ground heat exchanger. Conventional analysis methods employed by line source theory are used to predict heat transfer rate between ground heat exchanger and soil. Shape of ground heat exchanger was simplified by equivalent diameter model, but these methods do not accurately reflect the heat transfer characteristics according to the heat exchanger geometry. In this study, a numerical model that combines a user subroutine module that calculates circulation water conditions in the ground heat exchanger and FEFLOW program which can simulate heat/moisture transfer in the soil, is developed. Heat transfer performance was evaluated for 3 different types ground heat exchanger(U-tube, Double U-tube, Coaxial).

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2001.04a
• /
• pp.367-372
• /
• 2001

Feature restoration is that restore feature to 3D solid model using the feature information in STEP AP224. Feature is very important in CAPP, but feature information is defined very complicated in STEP AP224. This paper recommends the algorithm of extraction the feature information in physical STEP AP224file. This program import STEP AP224 file, parse the geometric and topological information, the tolerance data, and feature information line-by-line. After importation and parsing, store data into database. Feature restoration module analyze database including feature information, extract feature information, e.g. feature type, feature's parameter, etc., analyze the relationship and then restore feature to 3D solid model.

• Korean Journal of Computational Design and Engineering
• /
• v.5 no.4
• /
• pp.327-335
• /
• 2000

The spent fuel rod extracting machine is automatically operated in high radioactive environment, so high reliability of operation is required. In this paper, to enhance the reliability of this machine by providing a close monitoring capability. a real time graphic simulation method is suggested. This method utilizes conventional IGRIP (Interactive Graphics Robot Instruction Program) 3D graphic simulation tool to visualize and simulate the 3D graphic model of this machine. Also, the dedicated protocol is defined for transmission of the operational data of the machine. The real time graphic simulation is realized by developing the socket module between a graphic workstation and a machine control computer through the TCP/IP network and by dividing the 3D graphic simulation GSL(Graphic Simulation Language) program as a small sized sub routine. The suggested method is implemented while automatically operating the rod extracting machine. The result of implementation shows that the real time 3D graphic simulation is well synchronized with the actual machine according to the operational data.

• International journal of advanced smart convergence
• /
• v.9 no.2
• /
• pp.203-211
• /
• 2020

This study proposes a method for estimating gender and age that is robust to various external environment changes by applying deep learning-based learning. To improve the accuracy of the proposed algorithm, an improved CNN network structure and learning method are described, and the performance of the algorithm is also evaluated. In this study, in order to improve the learning method based on CNN composed of 6 layers of hidden layers, a network using GoogLeNet's inception module was constructed. As a result of the experiment, the age estimation accuracy of 5,328 images for the performance test of the age estimation method is about 85%, and the gender estimation accuracy is about 98%. It is expected that real-time age recognition will be possible beyond feature extraction of face images if studies on the construction of a larger data set, pre-processing methods, and various network structures and activation functions have been made to classify the age classes that are further subdivided according to age.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.83-86
• /
• 2002

This paper intends to provide convenience to the disabled, losing the use of their limbs, through voice recognition technology. The voice recognition part of this system recognizes voice by DTW (Dynamic Time Warping) Which is most Widely used in Speaker dependent system. Specially, S/N rate was improved through Wiener filter in the pre-treatment phase while considering real environmental conditions; the result values of 12th order feature pattern per frame are extracted by DTW algorithm using LPC and Cepsturm in feature extraction process. Furthermore, miniaturization is pursued using TMS320C32, 71's the floating-point DSP, for the hardware part. Currently, 90% of hardware porting has been completed, but we can confirm that the recognition rate was 96% as a result of performing the DTW algorithm in PC.