• Journal of Korean Neurosurgical Society
• /
• 제63권3호
• /
• pp.386-396
• /
• 2020

Objective : To generate synthetic spine magnetic resonance (MR) images from spine computed tomography (CT) using generative adversarial networks (GANs), as well as to determine the similarities between synthesized and real MR images. Methods : GANs were trained to transform spine CT image slices into spine magnetic resonance T2 weighted (MRT2) axial image slices by combining adversarial loss and voxel-wise loss. Experiments were performed using 280 pairs of lumbar spine CT scans and MRT2 images. The MRT2 images were then synthesized from 15 other spine CT scans. To evaluate whether the synthetic MR images were realistic, two radiologists, two spine surgeons, and two residents blindly classified the real and synthetic MRT2 images. Two experienced radiologists then evaluated the similarities between subdivisions of the real and synthetic MRT2 images. Quantitative analysis of the synthetic MRT2 images was performed using the mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results : The mean overall similarity of the synthetic MRT2 images evaluated by radiologists was 80.2%. In the blind classification of the real MRT2 images, the failure rate ranged from 0% to 40%. The MAE value of each image ranged from 13.75 to 34.24 pixels (mean, 21.19 pixels), and the PSNR of each image ranged from 61.96 to 68.16 dB (mean, 64.92 dB). Conclusion : This was the first study to apply GANs to synthesize spine MR images from CT images. Despite the small dataset of 280 pairs, the synthetic MR images were relatively well implemented. Synthesis of medical images using GANs is a new paradigm of artificial intelligence application in medical imaging. We expect that synthesis of MR images from spine CT images using GANs will improve the diagnostic usefulness of CT. To better inform the clinical applications of this technique, further studies are needed involving a large dataset, a variety of pathologies, and other MR sequence of the lumbar spine.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• 제20권11호
• /
• pp.305-313
• /
• 2019

Recently, spatial information is being constructed actively based on the images obtained by drones. Because occlusion areas occur due to buildings as well as many obstacles, such as trees, pedestrians, and banners in the urban areas, an efficient way to resolve the problem is necessary. Instead of the traditional way, which replaces the occlusion area with other images obtained at different positions, various models based on deep learning were examined and compared. A comparison of a type of feature descriptor, HOG, to the machine learning-based SVM, deep learning-based DNN, CNN, and RNN showed that the CNN is used broadly to detect and classify objects. Until now, many studies have focused on the development and application of models so that it is impossible to select an optimal model. On the other hand, the upgrade of a deep learning-based detection and classification technique is expected because many researchers have attempted to upgrade the accuracy of the model as well as reduce the computation time. In that case, the procedures for generating spatial information will be changed to detect the occlusion area and replace it with simulated images automatically, and the efficiency of time, cost, and workforce will also be improved.

• KIPS Transactions on Software and Data Engineering
• /
• 제3권12호
• /
• pp.511-522
• /
• 2014

During the development of the software, a variety of bugs are reported. Several bug tracking systems, such as, Bugzilla, MantisBT, Trac, JIRA, are used to deal with reported bug information in many open source development projects. Bug reports in bug tracking system would be triaged to manage bugs and determine developer who is responsible for resolving the bug report. As the size of the software is increasingly growing and bug reports tend to be duplicated, bug triage becomes more and more complex and difficult. In this paper, we present an approach to assign bug reports to appropriate developers, which is a main part of bug triage task. At first, words which have been included the resolved bug reports are classified according to each developer. Second, words in newly bug reports are selected. After first and second steps, vectors whose items are the selected words are generated. At the third step, TF-IDF(Term frequency - Inverse document frequency) of the each selected words are computed, which is the weight value of each vector item. Finally, the developers are recommended based on the similarity between the developer's word vector and the vector of new bug report. We conducted an experiment on Eclipse JDT and CDT project to show the applicability of the proposed approach. We also compared the proposed approach with an existing study which is based on machine learning. The experimental results show that the proposed approach is superior to existing method.

• Journal of Korean Neurosurgical Society
• /
• 제42권2호
• /
• pp.77-82
• /
• 2007

Objective : The authors tried to reveal some unique features of lipomeningomyelocele (LMMC), including clinical presentation, factors precipitating onset of symptoms, pathologic entities of LMMC associated with tethered cord syndrome, and surgical outcome in LMMC patients. Methods : Seventy-five patients with LMMC were enrolled in this study. Neuro-imaging and intraoperative findings allowed classification of LMMC into three Types. The patients were divided into two groups by age : A (51 patients), from birth to 3 years, and B (24 patients), from 3 to 24 years. For prevention of retethering of the cord, a mega-dural sac rebuilding procedure was performed in 15 patients. Results : During a mean postoperative follow-up period of 4 years, the surgical outcome was satisfactory in terms of improved pain and motor weakness, but disappointing with reference to the resolution of bowel and bladder dysfunction. Among these 75 patients with LMMC, preoperative deficits were improved after surgery in 29 (39%), remained stable in 28 (37%), changed slightly in 13 (17%), and worsened in 5 (7%). Patients in group A achieved better outcomes than those in group B. Depending on the type of lesion, patients with types I and II LMMC have better outcomes than those with type III LMMC. Finally, retethering of the cord with neurological deterioration occurred in 4 (5.3%) of the 75 patients, but no retethering was found in the 15 patients who were recently treated with a mega-dural sac rebuilding procedure. Conclusion : Our data continue to support the opinion that early diagnosis and optimal surgery are still essential for the treatment of patients with LMMC, since there is a high likelihood of residual neurological functions that can be preserved. Based on our surgical experience of untethering and decompression of lipomas, a mega-dural sac repair is useful to prevent retethering of the cord.

• Journal of KIISE
• /
• 제42권2호
• /
• pp.235-241
• /
• 2015

Locally linear embedding (LLE) [1] is a type of manifold algorithms, which preserves inner product value between high-dimensional data when embedding the high-dimensional data to low-dimensional space. LLE closely embeds data points on the same subspace in low-dimensional space, because the data points have significant inner product values. On the other hand, if the data points are located orthogonal to each other, these are separately embedded in low-dimensional space, even though they are in close proximity to each other in high-dimensional space. Meanwhile, it is well known that the facial images of the same person under varying illumination lie in a low-dimensional linear subspace [2]. In this study, we suggest an improved LLE method for face recognition problem. The method maximizes the characteristic of LLE, which embeds the data points totally separately when they are located orthogonal to each other. To accomplish this, all of the subspaces made by each class are forced to locate orthogonally. To make all of the subspaces orthogonal, the simultaneous Diagonalization (SD) technique was applied. From experimental results, the suggested method is shown to dramatically improve the embedding results and classification performance.

• Animal Systematics, Evolution and Diversity
• /
• 제15권2호
• /
• pp.153-164
• /
• 1999

One hundred and eleven samples of six subspecies of striped field mouse, Apodemus agrarius Pallas from Korea, China and Russia, were used for the analysis of mitochondrial DNA(mtDNA) fragment patterns resulted from the digestion with eight restriction enzymes by blot hybridization technique. All 32 fragments, nine mtDNA haplotypes, and four major subgroups with the mean divergence value of 0.896 to 1.150% were revealed. In summary, three forms are recognized: [I, subspecies chejuensis (Chejudo island, Korea)], [II, subspecies pallescens (southwestern Korea), coreae (central Korea), and septentrionalis (Russia)], and [III, subspecies manchuricus (northeastern China) and pallidior (northern China)], although some samples of subspecies coreae are somewhat different from almost all samples of six subspecies, and some samples of subspecies pallidior are similar with all samples of subspecies septentrionalis to form same haplotype. It is confirmed that A. agrarius chejuensis is a distinct subspecies, that subspecies coreae (including pallescens) is also a distinct subspecies, that subspecies manchuricus and pallidior are synonyms of subspecies ningpoensis, and that subspecies septentrionalis is a synonym of subspecies ningpoensis, and that subspecies septentrionalis is a synonym of subspecies agrarius. Moreover, it seems that A. agrarius shows constant karyotype, minimal variation in mtDNA genotype, and considerable divergence in morphometric characters, although further analyses with additional samples of A. agrarius in Eurasia will be necessary to determine the degree of variation of these taxonomic characters and to clarify subspecies classification as well.

• Journal of the Korean Association of Geographic Information Studies
• /
• 제2권3호
• /
• pp.1-15
• /
• 1999

This study was carried out to find out the way to build a comprehensive wetland ecosystem database using the technique of remote sensing and geographic information system. A Landsat TM image taken in May 17, 1997 was used for the primary source for the image analysis. Field surveys were conducted March to September of 1998 to help image analysis and examine the results. An actual wetland vegetation map was created based on the field survey. A Landsat TM image was analyzed by unsupervised and supervised classification methods and finally categorized into such 5 classes as Phragmites australis community, mixed community, sand beach, Scirpus trigueter community and non-vegetation intertidal area. Wetland basemap was developed for the overall accuracy assesment in wetland mapping. Vegetation index map of wetland vegetation was developed using NDVI(normalized difference vegetation index). The map of wetland productivity was developed based on the productivity of Phragmites australis and the relationship to the proximity of adjacent water bodies. The map of potential vegetation succession map was also developed based on the experience and knowledge of the field biologists. Considering these results, it is possible to use the remote sensing and GIS techniques for producing wetland ecosystem database. This study indicated that these techniques are very effective for the development of the national wetland inventory in Korea.

• Journal of the Institute of Electronics and Information Engineers
• /
• 제51권5호
• /
• pp.197-206
• /
• 2014

Motor imagery based Brain-computer Interface(BCI), which has recently attracted attention, is the technique for decoding the user's voluntary motor intention using Electroencephalography(EEG). For classifying the motor imagery, event-related desynchronization(ERD), which is the phenomenon of EEG voltage drop at sensorimotor area in ${\mu}$-band(8-13Hz), has been generally used but this method are not free from the performance degradation of the BCI system because EEG has low spatial resolution and shows different ERD-appearing band according to users. Common spatial pattern(CSP) was proposed to solve the low spatial resolution problem but it has a disadvantage of being very sensitive to frequency-band selection. Discriminative filter bank common spatial pattern(DFBCSP) tried to solve the frequency-band selection problem by using the Fisher ratio of the averaged EEG signal power and establishing discriminative filter bank(DFB) which only includes the feature frequency-band. However, we found that DFB might not include the proper filters showing the spatial pattern of ERD. To solve this problem, we apply a band-selection process using CSP feature vectors and linear discriminant analysis to DFBCSP instead of the averaged EEG signal power. The filter selection results and the classification accuracies of the existing and the proposed methods show that the CSP feature is more effective than signal power feature.

• 한국지구물리탐사학회:학술대회논문집
• /
• 한국지구물리탐사학회 2005년도 공동학술대회 논문집
• /
• pp.157-164
• /
• 2005

Urban conditions such as underground facilities and ambient noises due to cultural activity restrict the application of conventional geophysical techniques in general. We used the refraction microtremor (REMI) technique as an alternative way to get the geotechnical information, in particular shear-wave (S-wave) velocity information, at a site along an existing rail road. The REMI method uses ambient noises recorded using standard refraction equipment to derived shear-wave velocity information at a site. It does a wavefield transformation on the recorded wavefield to produce Rayleigh wave dispersion curve, which are then picked and modeled to get the shear-wave velocity structure. At this site the vibrations from the running trains provided strong noise sources that allowed REMI to be very effective. REMI was performed along the planned new underground rail tunnel. In addition, Suspension PS logging (SPS) were carried out at selected boreholes along the profile in order to draw out the quantitative relation between the shear wave velocity from the PS logging and the rock mass rating (RMR) determined from the inspection of the cores recovered from the same boreholes, These correlations were then used to relate the shear-wave velocity derived from REMI to RMR along the entire profile. The correlation between shear wave velocity and RMR was very good and so it was possible to estimate the RMR of the total zone of interest for the design of underground tunnel,

• Geophysics and Geophysical Exploration
• /
• 제9권1호
• /
• pp.108-113
• /
• 2006

Urban conditions, such as existing underground facilities and ambient noise due to cultural activity, restrict the general application of conventional geophysical techniques. At a tunnelling site in an urban area along an existing railroad, we used the refraction microtremor (REMI) technique (Louie, 2001) as an alternative way to get geotechnical information. The REMI method uses ambient noise recorded by standard refraction equipment and a linear geophone array to derive a shear-wave velocity profile. In the inversion procedure, the Rayleigh wave dispersion curve is picked from a wavefield transformation, and iteratively modelled to get the S-wave velocity structure. The REMI survey was carried out along the line of the planned railway tunnel. At this site vibrations from trains and cars provided strong seismic sources that allowed REMI to be very effective. The objective of the survey was to evaluate the rock mass rating (RMR), using shear-wave velocity information from REMI. First, the relation between uniaxial compressive strength, which is a component of the RMR, and shear-wave velocity from laboratory tests was studied to learn whether shear-wave velocity and RMR are closely related. Then Suspension PS (SPS) logging was performed in selected boreholes along the profile, in order to draw out the quantitative relation between the shear-wave velocity from SPS logging and the RMR determined from inspection of core from the same boreholes. In these tests, shear-wave velocity showed fairly good correlation with RMR. A good relation between shear-wave velocity from REMI and RMR could be obtained, so it is possible to estimate the RMR of the entire profile for use in design of the underground tunnel.