• Radiation Oncology Journal
• /
• v.25 no.4
• /
• pp.249-260
• /
• 2007

Purpose: In order to enhance the quality of IMRT as employed in Korea, we developed a remote monitoring system. The feasibility of the system was evaluated by conducting a pilot study. Materials and Methods: The remote monitoring system consisted of a head and neck phantom and a user manual. The phantom contains a target and three OARs (organs at risk) that can be detected on CT images. TLD capsules were inserted at the center of the target and at the OARs. Two film slits for GafchromicEBT film were located on the axial and saggital planes. The user manual contained an IMRT planning guide and instructions for IMRT planning and the delivery process. After the manual and phantom were sent to four institutions, IMRT was planed and delivered. Predicted doses were compared with measured doses. Dose distribution along the two straight lines that intersected at the center of the axial film was measured and compared with the profiles predicted by the plan. Results: The measurements at the target agreed with the predicted dose within a 3% deviation. Doses at the OARs that represented the thyroid glands showed larger deviations (minimum 3.3% and maximum 19.8%). The deviation at OARs that represented the spiral cord was $0.7{\sim}1.4%$. The percentage of dose distributions that showed more than a 5% of deviation on the lines was $7{\sim}27%$ and $7{\sim}14%$ along the horizontal and vertical lines, respectively. Conculsion: Remote monitoring of IMRT using the developed system was feasible. With remote monitoring, the deviation at the target is expected to be small while the deviation at the OARs can be very large. Therefore, a method that is able to investigate the cause of a large deviation needs to be developed. In addition, a more clinically relevant measure for the two-dimensional dose comparison and pass/fail criteria need to be further developed.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.3 no.2
• /
• pp.59-70
• /
• 1998

Two-dimensional trend-vector model of sediment transport is first tested in the tidal flat of Garolim Bay, mid-western coast of the Korean Peninsula. Three major parameters of surface sediment, i.e., mean grain size, sorting and skewness, are used for defining the best-fitting transport trend-vector on the sand ridge and muddy sand flat. These trend vectors are compared with the real transport directions determined from morphology, field observation and bedforms. The 15 possible cases of trend vectors are calculated from total sediments. In order to find the role of coarse sediments, trend vectors from sediments coarser than < 4.5 ${\phi}$, (sand size) are separately calculated from those of total sediments. As compared with the real directions, the best-fitting transport-vector model is the "case M" of coarse sediments which is the combined trend vectors of two cases: (1) finer, better sorted and more negatively skewed and (2) coarser, better sorted and more positively skewed. This indicates sand-size grains are formed by simpler hydrodynamic processes than total sediments. Transported sediment grains are better sorted than the source sediment grains. This indicates that consistent hydrodynamic energy can make sediment grains better sorted, regardless of complicated mechanisms of sediment transport. Consequently, both transported vector model and real transported direction show that the source of sediments are located outside of bay (offshore Yellow Sea) and in the baymouth. These source sediments are transported through the East Main Tidal Channel adjacent the baymouth. Some are transported from the subtidal zone to the upper tidal flat, but others are transported farther to the south, reaching the south tidal channel in the study area. Also, coarse sediment grains on the sand ridge are originally from the baymouth, and transported through the subtidal zone to the south tidal channel. These coarse sediments are moved to the northeast, but could not pass the small north tidal channel. It is interpreted that the great amount of coarse sediments is returned back to the outside of the bay (Yellow Sea) again through the baymouth during the ebb tide. The distribution of muddy sand in the northeastern part of study area may result from the mixing of two sediment transport mechanisms, i.e., suspension and bedload processes. The landward movement of sand ridge and the formation of the north tidal channel are formed either by the supply of coarse sediments originating from the baymouth and outside of the bay (subaqueous sand ridges including Jang-An-Tae) or by the recent relative sea-level rise.

• Korean Journal of Crystallography
• /
• v.6 no.2
• /
• pp.125-133
• /
• 1995

The crystal structure of dehydrated, partially Co(II)-exchanged zeolite X, stoichiometry Co2+Na+-X (Co41+Na10Si100Al92O384) per unit cell, has been determined from three-dimensional X-ray diffraction data gathered by counter methods. The structure was solved and refined in the cubic space group Fd3:α=24.544(1)Å at 21(1)℃. The crystal was prepared by ion exchange in a flowing stream using a solution 0.025 M each in Co(NO3)2 and Co(O2CCH3)2. The crystal was then dehydrated at 380℃ and 2×10-6 Torr for two days. The structure was refined to the final error indices, R1=0.059 and R2=0.046 with 211 reflections for which I > 3σ(I). Co2+ ions and Na+ ions are located at the four different crystallographic sites. Co2+ ions are located at two different sites of high occupancies. Sixteen Co2+ ions are located at the center of the double six-ring (site I; Co-O = 2.21(1)Å, O-Co-O = 90.0(4)°) and twenty-five Co2+ ions are located at site II in the supercage. Twenty-five Co2+ ions are recessed 0.09Å into the supercage from its three oxygen plane (Co-O = 2.05(1)Å, O-Co-O = 119.8(7)°). Na+ ions are located at two different sites of occupandies. Seven Na+ ions are located at site II in the supercage (Na-O = 2.29(1)Å, O-Na-O = 102(1)°). Three Na+ ions are statistically distribyted over site III, a 48-fold equipoint in the supercages on twofold axes (Na-O = 2.59(10)Å, O-Na-O = 69.0(3)°). Seven Na+ ions are recessed 1.02Å into the supercage from the three oxygen plane. It appears that Co2+ ions prefer sites I and II in order, and that Na+ ions occupy the remaining sites, II and III.

• Journal of the Korean Vacuum Society
• /
• v.19 no.4
• /
• pp.307-313
• /
• 2010

One-dimensional cubic phase silicon carbide nanowires (${\beta}$-SiC NWs) were efficiently synthesized by thermal chemical vapor deposition (TCVD) with mixtures containing Si powders and nickel chloride hexahydrate $(NiCl_2{\cdot}6H_2O)$ in an alumina boat with a carbon source of methane $(CH_4)$ gas. SEM images are shown that the growth temperature (T) of $1,300^{\circ}C$ is not enough to synthesize the SiC NWs owing to insufficient thermal energy for melting down a Si powder and decomposing the methane gas. However, the SiC NWs could be synthesized at T>$1,300^{\circ}C$ and the most efficient temperature for growth of SiC NWs is T=$1,400^{\circ}C$. The synthesized SiC NWs have the diameter with an average range between 50~150 nm. Raman spectra clearly revealed that the synthesized SiC NWs are forming of a cubic phase (${\beta}$-SiC). Two distinct peaks at 795 and $970 cm^{-1}$ in Raman spectra of the synthesized SiC NWs at T=$1,400^{\circ}C$ represent the TO and LO mode of the bulk ${\beta}$-SiC, respectively. XRD spectra are also supported to the Raman spectra resulting in the strongest (111) peaks at $2{\Theta}=35.7^{\circ}$, which is the (111) plane peak position of 3C-SiC. Moreover, the gas flow rate of 300 sccm for methane is the optimal condition for synthesis of a large amount of ${\beta}$-SiC NW without producing the amorphous carbon structure shown at a high methane flow rate of 800 sccm. TEM images are shown two kinds of the synthesized ${\beta}$-SiC NWs structures. One is shown the defect-free ${\beta}$-SiC NWs with a (111) interplane distance of 0.25 nm, and the other is the stacking-faulted ${\beta}$-SiC NWs. Also, TEM images exhibited that two distinct SiC NWs are uniformly covered with $SiO_2$ layer with a thickness of less 2 nm.

• The Journal of Korean Academy of Prosthodontics
• /
• v.57 no.3
• /
• pp.203-210
• /
• 2019

Purpose: The purpose of this study was to verify the effect of the abutment superimposition process on the final virtual model in the scanning process of single and 3-units bridge model using a dental model scanner. Materials and methods: A gypsum model for single and 3-unit bridges was manufactured for evaluating. And working casts with removable dies were made using Pindex system. A dental model scanner (3Shape E1 scanner) was used to obtain CAD reference model (CRM) and CAD test model (CTM). The CRM was scanned without removing after dividing the abutments in the working cast. Then, CTM was scanned with separated from the divided abutments and superimposed on the CRM (n=20). Finally, three-dimensional analysis software (Geomagic control X) was used to analyze the root mean square (RMS) and Mann-Whitney U test was used for statistical analysis (${\alpha}=.05$). Results: The RMS mean abutment for single full crown preparation was $10.93{\mu}m$ and the RMS average abutment for 3 unit bridge preparation was $6.9{\mu}m$. The RMS mean of the two groups showed statistically significant differences (P<.001). In addition, errors of positive and negative of two groups averaged $9.83{\mu}m$, $-6.79{\mu}m$ and 3-units bridge abutment $6.22{\mu}m$, $-3.3{\mu}m$, respectively. The mean values of the errors of positive and negative of two groups were all statistically significantly lower in 3-unit bridge abutments (P<.001). Conclusion: Although the number of abutments increased during the scan process of the working cast with removable dies, the error due to the superimposition of abutments did not increase. There was also a significantly higher error in single abutments, but within the range of clinically acceptable scan accuracy.

• Korean Journal of Remote Sensing
• /
• v.37 no.3
• /
• pp.419-430
• /
• 2021

Remote sensing techniques using drone-based multispectral image were studied for fast and two-dimensional monitoring of algal blooms in the river. Drone is anticipated to be useful for algal bloom monitoring because of easy access to the field, high spatial resolution, and lowering atmospheric light scattering. In addition, application of multispectral sensors could make image processing and analysis procedures simple, fast, and standardized. Spectral indices derived from the active spectrum of photosynthetic pigments in terrestrial plants and phytoplankton were tested for estimating chlorophyll-a concentrations (Chl-a conc.) from drone-based multispectral image. Spectral indices containing the red-edge band showed high relationships with Chl-a conc. and especially, 3-band model (3BM) and normalized difference chlorophyll index (NDCI) were performed well (R²=0.86, RMSE=7.5). NDCI uses just two spectral bands, red and red-edge, and provides normalized values, so that data processing becomes simple and rapid. The 3BM which was tuned for accurate prediction of Chl-a conc. in productive water bodies adopts originally two spectral bands in the red-edge range, 720 and 760 nm, but here, the near-infrared band replaced the longer red-edge band because the multispectral sensor in this study had only one shorter red-edge band. This index is expected to predict more accurately Chl-a conc. using the sensor specialized with the red-edge range.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.95-108
• /
• 2017

Recently, AlphaGo which is Bakuk (Go) artificial intelligence program by Google DeepMind, had a huge victory against Lee Sedol. Many people thought that machines would not be able to win a man in Go games because the number of paths to make a one move is more than the number of atoms in the universe unlike chess, but the result was the opposite to what people predicted. After the match, artificial intelligence technology was focused as a core technology of the fourth industrial revolution and attracted attentions from various application domains. Especially, deep learning technique have been attracted as a core artificial intelligence technology used in the AlphaGo algorithm. The deep learning technique is already being applied to many problems. Especially, it shows good performance in image recognition field. In addition, it shows good performance in high dimensional data area such as voice, image and natural language, which was difficult to get good performance using existing machine learning techniques. However, in contrast, it is difficult to find deep leaning researches on traditional business data and structured data analysis. In this study, we tried to find out whether the deep learning techniques have been studied so far can be used not only for the recognition of high dimensional data but also for the binary classification problem of traditional business data analysis such as customer churn analysis, marketing response prediction, and default prediction. And we compare the performance of the deep learning techniques with that of traditional artificial neural network models. The experimental data in the paper is the telemarketing response data of a bank in Portugal. It has input variables such as age, occupation, loan status, and the number of previous telemarketing and has a binary target variable that records whether the customer intends to open an account or not. In this study, to evaluate the possibility of utilization of deep learning algorithms and techniques in binary classification problem, we compared the performance of various models using CNN, LSTM algorithm and dropout, which are widely used algorithms and techniques in deep learning, with that of MLP models which is a traditional artificial neural network model. However, since all the network design alternatives can not be tested due to the nature of the artificial neural network, the experiment was conducted based on restricted settings on the number of hidden layers, the number of neurons in the hidden layer, the number of output data (filters), and the application conditions of the dropout technique. The F1 Score was used to evaluate the performance of models to show how well the models work to classify the interesting class instead of the overall accuracy. The detail methods for applying each deep learning technique in the experiment is as follows. The CNN algorithm is a method that reads adjacent values from a specific value and recognizes the features, but it does not matter how close the distance of each business data field is because each field is usually independent. In this experiment, we set the filter size of the CNN algorithm as the number of fields to learn the whole characteristics of the data at once, and added a hidden layer to make decision based on the additional features. For the model having two LSTM layers, the input direction of the second layer is put in reversed position with first layer in order to reduce the influence from the position of each field. In the case of the dropout technique, we set the neurons to disappear with a probability of 0.5 for each hidden layer. The experimental results show that the predicted model with the highest F1 score was the CNN model using the dropout technique, and the next best model was the MLP model with two hidden layers using the dropout technique. In this study, we were able to get some findings as the experiment had proceeded. First, models using dropout techniques have a slightly more conservative prediction than those without dropout techniques, and it generally shows better performance in classification. Second, CNN models show better classification performance than MLP models. This is interesting because it has shown good performance in binary classification problems which it rarely have been applied to, as well as in the fields where it's effectiveness has been proven. Third, the LSTM algorithm seems to be unsuitable for binary classification problems because the training time is too long compared to the performance improvement. From these results, we can confirm that some of the deep learning algorithms can be applied to solve business binary classification problems.

• Journal of KIISE:Databases
• /
• v.30 no.6
• /
• pp.651-665
• /
• 2003

To reduce cache misses emerges as the most important issue in today's situation of main memory databases, in which CPU speeds have been increasing at 60% per year, and memory speeds at 10% per year. Recent researches have demonstrated that cache-conscious index structure such as the CR-tree outperforms the R-tree variants. Its search performance can be poor than the original R-tree, however, since it uses a lossy compression scheme. In this paper, we propose alternatively a cache-conscious version of the R-tree, which we call MR-tree. The MR-tree propagates node splits upward only if one of the internal nodes on the insertion path has empty room. Thus, the internal nodes of the MR-tree are almost 100% full. In case there is no empty room on the insertion path, a newly-created leaf simply becomes a child of the split leaf. The height of the MR-tree increases according to the sequence of inserting objects. Thus, the HeightBalance algorithm is executed when unbalanced heights of child nodes are detected. Additionally, we also propose the CCMR-tree in order to build a more cache-conscious MR-tree. Our experimental and analytical study shows that the two-dimensional MR-tree performs search up to 2.4times faster than the ordinary R-tree while maintaining slightly better update performance and using similar memory space.

• Journal of Korean Academy of Oral and Maxillofacial Radiology
• /
• v.29 no.2
• /
• pp.423-433
• /
• 1999

Purpose : To assess the methods for the clinical evaluation of the longitudinal bone changes after implantation of tooth ash-plaster mixture into the defect area of human jaws. Materials and methods : Tooth ash-plaster mixtures were implanted into the defects of 8 human jaws. 48 intraoral radiograms taken with copper step wedge as reference at soon, 1st, 2nd, 4th, and 6th week after implantation of mixture were used. X-ray taking was standardized by using Rinn XCP device customized directly to the individual dentition with resin bite block. The images inputted by Quick scanner were digitized and analyzed by NIH image program. Cu­equivalent values were measured at the implanted sites from the periodic digital images. Analysis was performed by the bidirectional subtraction with color enhancement and the surface plot of resliced contiguous image. The obtained results by the two methods were compared with Cu­equivalent value changes. Results : The average determination coefficient of Cu-equivalent equations was 0.9988 and the coefficient of variation of measured Cu values ranged from 0.08~0.10. The coefficient of variation of Cu-equivalent values measured at the areas of the mixture and the bone by the conversion equation ranged from 0.06 ~0.09. The analyzed results by the bidirectional subtraction with color enhancement were coincident with the changes of Cu-equivalent values. The surface plot of the resliced contiguous image showed the three dimensional view of the longitudinal bone changes on one image and also coincident with Cu-equivalent value changes after implantation. Conclusion : The bidirectional subtraction with color enhancement and the surface plot of the resliced contiguous image was very effective and reasonable to analyze clinically and qualitatively the longitudinal bone change. These methods are expected to be applicable to the non-destructive test in other fields.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.12
• /
• pp.1970-1982
• /
• 1993

This paper describes an efficient array algorithm for parallel computation of vector-radix two-dimensional (2-D) fast discrete cosine transform (VR-FCT), and its VLSI implementation. By mapping the 2-D VR-FCT onto a 2-D array of processing elements (PEs), the butterfly structure of the VR-FCT can be efficiently importanted with high concurrency and local communication geometry. The proposed array algorithm features architectural modularity, regularity and locality, so that it is very suitable for VLSI realization. Also, no transposition memory is required, which is invitable in the conventional row-column decomposition approach. It has the time complexity of O(N+Nnzp-log2N) for (N*N) 2-D DCT, where Nnzd is the number of non-zero digits in canonic-signed digit(CSD) code, By adopting the CSD arithmetic in circuit desine, the number of addition is reduced by about 30%, as compared to the 2`s complement arithmetic. The computational accuracy analysis for finite wordlength processing is presented. From simulation result, it is estimated that (8*8) 2-D DCT (with Nnzp=4) can be computed in about 0.88 sec at 50 MHz clock frequency, resulting in the throughput rate of about 72 Mega pixels per second.