• Radiation Oncology Journal
• /
• v.27 no.4
• /
• pp.194-200
• /
• 2009

Purpose: This aim of this study was to evaluate changes in gastric volume and organ position as a result of delayed gastric emptying after a subtotal gastrectomy performed as part of the treatment of stomach cancer. Materials and Methods: The medical records of 32 patients who underwent concurrent chemoradiotherapy after a subtotal gastrectomy from March 2005 to December 2008 were reviewed. Of these, 5 patients that had more than 50 cc of residual gastric food detected at computed tomography (CT) simulation, were retrospectively enrolled in this study. Gastric volume and organ location was measured from CT images obtained before radiotherapy, twice weekly. In addition, authors evaluated the change of radiation dose distribution to planning the target volume and normal organ in a constant radiation therapy plan regardless of gastric volume variation. Results: A variation in the gastric volume was observed during the radiotherapy period (64.2~340.8 cc; mean, 188.2 cc). According to the change in gastric volume, the location of the left kidney was shifted up to 0.7 - 2.2 cm (mean, 1.2 cm) in the z-axis. Under-dose to planning target volume (V43, 79.5${\pm}$10.4%) and over-dose to left kidney (V20, 34.1${\pm}$12.1%; Mean dose, 23.5${\pm}$8.3 Gy) was expected, given that gastric volume change due to delayed gastric emptying wasn't taken into account. Conclusion: This study has shown that a great change in gastric volume and left kidney location may occur during the radiation therapy period following a subtotal gastrectomy, as a result of delayed gastric emptying. Detection of patients who experienced delayed gastric emptying and the application of gastric volume variation to radiation therapy planning will be very important.

• The Journal of Korean Medical History
• /
• v.14 no.2
• /
• pp.109-152
• /
• 2001

Throughout this paper, I adjusted the study of 'Qian Yi'(錢乙)'s Medical Thought, and the following is the summary. 1. 'Qian Yi' wrote 'Xiao Er Yao Zheng Zhi Jue'("小兒藥證直訣", edited by 誾季忠), and there were 'Shang Han Lun Zhi Wei'("傷寒論指微"), 'Ying Ru Lun', however those are loss of the record. 2. Qian Yi's 'Zhi Jue'("直訣") was influenced by 'Lu Xin Jing', yet if we compare the quality of 'Sheng Li, Byeng Li, Bang Jae'(生理, 病理, 方劑), 'Lu Xin Jing' cannot be the foundation of 'Zhi Jue'. He took over 'Nei Jing, Shang Han Lun, Jin Gui Yao Lue, Shen Long Ben Cao Jing, Tai Ping Sheng Hui Fang'("內經", "傷寒論", "金?要略", "神膿本草經", "太平聖惠方") and put them together to the direct clinical experiences of pediatrics. 3. There is no reference regarding the difficulties of pediatric diagnosis and diseases in 'Huang Di Nei Jing'("黃帝內經") Before 'Bei Song'(北宋), regardless of the lack of data related to pediatric diseases, 'Qian Yi' established the pediatric system in 'Xiao Er Yao Zheng Zhi Jue' for the first time. 4. In his diagnosis of the pediatric diseases, he 'Si Zhen He Can'(四診合參), also considered in the eye exam seriously. In addition, he closely combined 'Wu Zang Bian Zheng'(五臟辨證), and diagnosis the pediatric diseases. 5. 'Wu Zang Bian Zheng', what Qian established method was based on 'Zheng Ti Guan'(整體觀) in 'Huang Di Nei Jing'. It was based on clinical experiences and established the perspectives of 'Tian Ren Xiang Ying'(天人相應). First of all, he pinpointed 'Zhu Zheng'(主證) clearly. Secondly, he pinpointed the relationships to symptoms and then, he distinguished a generic character of 'Xu, Shi, Han, Re'(虛, 實, 寒, 熱). Finally, he made an induction from genealogical pediatric physiology. 6. 'Qian Yi' took a serious view of 'Ban Zhen'(斑疹), the inadequate field in those days. At that time, he criticized on the habituation of the misuse of medication. He treated separately which 'Ji Jing'(急驚) as 'Liang Xie'(凉瀉) and 'Man Jing'(慢驚) as 'Wen Bu'(溫補). He proposed 'Cong Gan Zhu Feng, Xin Zhu Jing'(從肝主風, 心主驚) theory and formulated 'Jing Feng'(驚風) theory as well. 7. As an opponent of a tendency to misusage of medicine, 'Qian Yi' made out a prescription with pliant medicine. He emphasized on the treatment to 'Gong Bu Shang Zheng, Bu Bu Zhi Xie, Xiao Bu Jian Shi'(攻不傷正, 補不滯邪, 消補兼施) because he had so lucid demonstration to 'Xu Shi Han Re'(虛實寒熱) of the five viscera in the field of 'Bang Yak'(方藥). 8. There were no pediatrics schools at that time, however, the pediatrics was being made up gradually by 'Jin Yuan Si Da Jia'(金元四大家) who was influenced by 'Qian Yi'. He raised an objection to medical treatment using pliant medicine. 'Qian Yi' applied 'Qu Xia'(驅下) treatment using 'Han Liang'(寒凉) medicine. 'Han Liang Pai'(寒凉派) is greatly influenced by Qian. 'Chen Wen Zhong'(陳文中) had a great impact on 'Han Liang Pai' who used a 'Zao Shu Wen Bu'(燥熟溫補) medicine for treatment. Since 'Song Jin'(宋金), he had a tremendous influence on pediatrics treating patients in both 'Han Wen'(寒溫) ways. 9. 'Qian Yi' had an influence on his medical thoughts on future generations, especially to 'Wan Quan'(萬全) of 'Ming Dai', 'Wu Tang'(吳塘) of 'Qing Dai'(淸代) and 'Yun Shu Jie'(?樹珏) of 'Min Guo'(民國). 'Wan Quan' is an advocate of 'You Yu, Bu Zu Zhi Shuo'(有餘, 不足之說)of 'Xiao Er Wu Zang'(小兒五臟) that he revealed Qian's 'Wu Zang Bian Zheng'(五臟辨證). 'Wu Tang' disclosed Qian's 'Xiao Er Ti Zhi Shuo'(小兒體質說) and 'Xiao Er Ke'(小兒科)'s 'Yong Yao Lun'(用藥論), therefore, he uncovered pediatric physiological characteristics through the advocate of Qian's 'Zang Fu Rou Ruo, Ji Gu Nen Qie, Yi Xu Yi Shi, Yi Han Yi Re' (臟腑柔弱, 肌骨嫩怯, 易虛易實, 易寒易熱). 'Yun Shu Jie' developed intrinsic relationships among time, symptom and 'Tian Ren Xiang Ying Guan'(天人相應觀), What 'Qian Yi' stated about them. And also, he developed Qian's 'Di Huang Wan'(地黃丸), 'Xie Qing Wan'(瀉靑丸), 'Yi Huang San'(益黃散) clinical usages as well. 10. Regarding Qian's 'Wu Zang Xu Shi'(五臟虛實), it has an influence on 'Zhang Yuan Su'(張元素)'s 'Zang Fu Bing Ji Bian Zheng'(臟腑病機辨證). 'Di Huang Wan', 'Xie Qing Wan', 'Xie Xin Tang'(瀉心湯), 'Yi Huang San', 'Xie Huang San'(瀉黃散) are the standard prescription of 'Wu Zang Bu Xie'(五臟補瀉). It is under the influence of Qian's treatment. Besides, 'Qian Yi' took a serious view of 'Xiao Er'(小兒)'s 'Pi Wei'(脾胃). 'Qian Yi' had an impact on 'Li Dong Yuan'(李東垣) one of the member of 'Bu Tu Pai'(補土派). 'Di Huang Wan', which placed great importance on 'Bu Yi Shen Yin'(補益腎陰), had a great impact on 'Da Bu Yin Wan'(大補陰丸) and 'Jin Yuan Si Da Jia' as well. 11. In a theory of Qian's 'Wu Zang Bian Zheng', though it had been stated clearly in 'Wu Zang Bian Zheng', but he neglected in 'Liu Fu Bian Zheng'(六腑辨證). In prescription field, The problem with the medicine is that it is either toxic or mineral, therefore, we are not able to use those medicine in a clinical testing at the present time.

• Korean Journal of Soil Science and Fertilizer
• /
• v.34 no.5
• /
• pp.326-332
• /
• 2001

To investigate a fertility status of upland soil, the soil were sampled at 854 sites chosen in consideration of areal distribution percent on the basis of topography and were analyzed on pH, organic mater, available phosphorus ($P_2O_5$), exchangeable potassium and calcium and magnesium. The content of soil chemical properties showed pH 5.6, organic mater $24g\;kg^{-1}$, available $P_2O_5$ $577mg\;kg^{-1}$, exchangeable potassium and calcium and magnesium were 0.85, 4.5, $1.4cmol^+kg^{-1}$, respectively. The distribution percent of soil samples within the optimum range for cropping were 13.4% for pH, 46.7% for organic matter. 27.4% for available $P_2O_5$, 10.7, 15.8, 18.3% for exchangeable potassium and calcium and magnesium, respectively. In chronological changes of soil properties, exchangeable calcium and magnesium were ignorable ; pH was slightly decreased ; organic matter was slightly increased ; available $P_2O_5$ and exchangeable potassium were greatly increased.

• Korean Journal of Heritage: History & Science
• /
• v.52 no.4
• /
• pp.56-77
• /
• 2019

Bi-Seok is a treasure trove of funeral rites and an important cultural asset that can shed light on the historical and social history of calligraphy, but research of the topic is still insignificant. In particular, research on the production method of Bi-Seok remains an unproven field. The production of Bi-Seok can be roughly divided into ma-jeong (refining stone), sculpture, and the Buk-chil (process of engraving letters) process. This article reveals some facts: First, performing ma-jeong to the Sang-Seok, Honyu-Seok, Bi-seok, which are known to be God's things. This process is needed because of the change in the perception of the Honyu-Seok due to the settlement and propagation of Confucian ceremonial rituals in the times of hardship in 1592 and 1636. As the crafting process of ma-jeong did not remain concrete, it was only possible to examine the manufacturing process of Bi-Seok through its materials and tools. Second, the rapid proliferation of Oh-Seok and Sa-jeo-chwi-yong (purchase of things made by private citizens) in the Yeongjo era has great importance in social and cultural history. When the Gang-Hwa-Seok of the commodity were exhausted, the Oh-Seok that was used by Sadebu (upper civil class) were used in the tomb of Jangneung, which made Oh-Seok popular among people. In particular, the use of Oh-Seok and the Ma-Jeong process could minimize chemical and physical damage. Third, the writing method of the Bi-seok is Buk-chil. After Buk-Chil of Song Si-Yeol was used on King Hyojong's tomb, the Buk-Chil process ( printing the letters on the back of the stone and rubbing them to make letters) became the most popular method in Korea and among other East Asian countries, and the fact that it was institutionalized to this scale was quite impressive. Buk-Chil became more sophisticated by using red ink rather than black ink due to the black color that results from Oh-Seok. Fourth, the writing method changes in the late Joseon Dynasty. Until the time of Yeongjo's regime, when inscribing, the depth of the angle was based on the thickness of the stroke, thus representing the shade. This technique, of course, did not occur at every Pyo-Seok or Shindobi, but was maintained by outstanding artisans belonging to government agencies. Therefore, in order to manufacture Bi-Seok, Suk-seok, YeonJeong, Ma-jeong, Jeong-Gan, ChodoSeoIp, Jung-Cho, Ip-gak, Gyo-Jeong, and Jang-Hwang, a process was needed to make one final product. Although all of these methods serve the same purpose of paying respects and propagandizing the great work of deceased persons, through this analysis, it was possible to see the whole process of Pyo-Seok based upon the division of techniques and the collaboration of the craftsmen.

• Journal of Veterinary Clinics
• /
• v.24 no.1
• /
• pp.10-14
• /
• 2007

The present study was performed to elucidate the recovery effect of needle-acupuncture (needle-AP) and moxibustion treatments at Gan Shu (BL18) on $CCl_4$ induced hepatic injury in dogs, Total 9 clinically healthy experimental dogs (1 to 2 years old, 2.3-5.3kg body weight) were divided into control (3 dogs), needle-AP (3 dogs) and moxibustion (3 dogs) groups, respectively. Hepatic iniw was induced by intraperitoneal injection with $CCl_4$ (olive oil : $CCl_4=1:1,\;1ml/kg$, once/day). As far the treatments in each group, control group was not treated at all after induction of the hepatic injury till the end of experiment. Needle-Af group was treated by AP at BL18 (once/day, 20 minutes) from the next day of induction of the hepatic injury for 11 days. Moxibustion group was also treated with commercial moxa at BL18 (once/day, 20minutes) for 11days. The changes of the serum alanine transaminase(ALT), aspartate transaminase(AST) and gamma glutamyl transpeptidase (GGT) activities were investigated on pre, 0, 1, 3, 5, 7 and 11 days after hepatic injury, respectively. Histopathological changes were also investigated in the liver tissues on day 11 in experimental and control groups. The results obtained in the present study were as follows. In needle-AP group, they showed significant lower values on the 1st (p<0.05) and $3^{rd} day (p <0.05)$ in serum ALT activities, and the $11^{th}$ day in serum GGT activities, compared with those of control group, respectively. However, the significances were not detected in serum AST activities of needle-AP group by comparison with those of control group. In moxibustion group, they showed significant low values on the $1^{st}(p<0.05)$, 3rd day (p<0.05) and $5^{th}$ day (p<0.05) in serum ALT activities, compared with those of control group, respectively. However, significances were not detected in serum GGT and AST activities of moxibustion group by comparison with those of control group. Among the experimental groups, the significant low values were found on $1^{st}\;and\;3^{rd}$ day in serum ALT activities (p<0.05), and $3^{rd}\;and\;5^{th}$ day in serum GGT activities (p<0.05) in needle-AP group, compared by those of moxibustion group, respectively. In addition, the marked recovery findings of histopathological changes in experimental groups were found, compared by those of control group: mild vacuolar degeneration and necrosis findings in needle-AP group, and moderate vacuolar degeneration and necrotic findings in moxibustion group were found, in contrast to severe changes such as accumulation of bile juice, vacuolar degeneration and necrotic findings observed in control group. In conclusion, needle-AP and moxibustion treatments at BL-18 were effective for the recovery for $CCl_4$ induced hepatic injury and needle-AP was more effective than that of moxibustion in dogs.

• Korean Journal of Soil Science and Fertilizer
• /
• v.41 no.2
• /
• pp.75-82
• /
• 2008

In order to establish environmentally friendly phosphorus(P) and potassium(K) fertilizer application rates for rice cultivation in Korea, a field experiment was conducted in two different characterized paddy soils, higher and lower P and K content in soil from 2005 to 2006 in Suwon, Korea. Chuchong which is typical rice variety in Korea, was selected for experiment. Rice yields in base P-K fertilizer level ($30-30kg\;ha^{-1}$) and no P-K fertilizer level were 97~98% and 93~97% of standard P-K fertilizer level ($45-57kg\;ha^{-1}$), respectively. At the lower content of P and K in immatured paddy soil, the yield of no P plot was decreased by 7% compared to that of standard P-K plot. Significant difference in rice yield and yield characteristics was found neither in standard P-K fertilization plot nor in base P-K fertilization plot. There was no significant difference in P and K uptake at the harvest stage between standard P-K fertilizer plot and base P-K fertilizer plot. Phosphorus and K uptake was slightly decreased in no fertilizer plots compared to standard fertilizer plot. Phosphorus and K use efficiency of rice plant in base fertilizer plot was higher than that in standard fertilizer plot. Available phosphorus and exchangeable potassium content in soil at the harvest stage were increased in both standard and base fertilizer plot. But no significant difference of P and K content in ? was found between standard fertilizer plot and base fertilizer plot. Rice quality characteristics such as proportion of brewers rice, damaged kernel, broken rice, and content of amylose and protein were lowered in the base fertilizer plot than the standard fertilized plot, while the proportion of perfect kernel increased in the base fertilizer plot From the results, we generally recommend the base P-K fertilization ($30-30kg\;ha^{-1}$) instead of standard P-K fertilization ($45-57kg\;ha^{-1}$) for rice cultivation in Korea.

• Korean Journal of Soil Science and Fertilizer
• /
• v.40 no.5
• /
• pp.354-363
• /
• 2007

In order to investigate the optimum rate and interval of silicate fertilizer application for rice cultivation, Chucheong byeo variety, one of commonly cultivated rice cultivar in Korea was planted on two different wetland rice soils located on Hwaseong-si from 2002 to 2005; Jisan series(a member of the fine loamy, mixed, mesic family of Fluvaquentic Endoaquepts), known as "Productive Paddy Soil", without any conspicuous limiting factor, and Seokcheon series (a member of the coarse loamy, mixed, nonacid, mesic family of Fluvaquentic Endoaquetps), known as "Sandy Paddy Soil", sandiness being major limiting factor. There were three rate treatments of silicate fertilizer application; the amount of silicate fertilizers needed to adjust the available soil silicate contents to 130, 200, and $270mg\;kg^{-1}$ was applied, in the first year only. There was an additional plot; applying the amount of silicate fertilizer needed to adjust soil available silicate to 130 ppm every year, which would serve as the base for the evaluation of residual effects of silicate fertilizers in the plots where different rates of silicate fertilizer were applied. From the yield data in first year, it was found that optimum available silica in the soil are $154mg\;kg^{-1$ and $160mg\;kg^{-1}$, in Jisan and Seogcheon soils, respectably. The duration of residual effects of silicate fertilizer was different depending upon the amount of applied silicate fertilizers and the soils. The higher the application rate, the residual effect lasted longer, and the residual effect was lasted longer in Jisan(clay loam) soil than in Seogcheon(sandy loam) soil. During four years, sum of the rate of contribution to increase available soil silica of applied silicate fertilizer in different soils ranged 18.6% and 24.1% in Jisan soil and Seogcheon soil, respectively. This may suggest that much portion of applied silicate would be either lost from the soil or remain in the soil as insoluble form. This deserves further study.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.43-62
• /
• 2019

At one time, the anomaly detection sector dominated the method of determining whether there was an abnormality based on the statistics derived from specific data. This methodology was possible because the dimension of the data was simple in the past, so the classical statistical method could work effectively. However, as the characteristics of data have changed complexly in the era of big data, it has become more difficult to accurately analyze and predict the data that occurs throughout the industry in the conventional way. Therefore, SVM and Decision Tree based supervised learning algorithms were used. However, there is peculiarity that supervised learning based model can only accurately predict the test data, when the number of classes is equal to the number of normal classes and most of the data generated in the industry has unbalanced data class. Therefore, the predicted results are not always valid when supervised learning model is applied. In order to overcome these drawbacks, many studies now use the unsupervised learning-based model that is not influenced by class distribution, such as autoencoder or generative adversarial networks. In this paper, we propose a method to detect anomalies using generative adversarial networks. AnoGAN, introduced in the study of Thomas et al (2017), is a classification model that performs abnormal detection of medical images. It was composed of a Convolution Neural Net and was used in the field of detection. On the other hand, sequencing data abnormality detection using generative adversarial network is a lack of research papers compared to image data. Of course, in Li et al (2018), a study by Li et al (LSTM), a type of recurrent neural network, has proposed a model to classify the abnormities of numerical sequence data, but it has not been used for categorical sequence data, as well as feature matching method applied by salans et al.(2016). So it suggests that there are a number of studies to be tried on in the ideal classification of sequence data through a generative adversarial Network. In order to learn the sequence data, the structure of the generative adversarial networks is composed of LSTM, and the 2 stacked-LSTM of the generator is composed of 32-dim hidden unit layers and 64-dim hidden unit layers. The LSTM of the discriminator consists of 64-dim hidden unit layer were used. In the process of deriving abnormal scores from existing paper of Anomaly Detection for Sequence data, entropy values of probability of actual data are used in the process of deriving abnormal scores. but in this paper, as mentioned earlier, abnormal scores have been derived by using feature matching techniques. In addition, the process of optimizing latent variables was designed with LSTM to improve model performance. The modified form of generative adversarial model was more accurate in all experiments than the autoencoder in terms of precision and was approximately 7% higher in accuracy. In terms of Robustness, Generative adversarial networks also performed better than autoencoder. Because generative adversarial networks can learn data distribution from real categorical sequence data, Unaffected by a single normal data. But autoencoder is not. Result of Robustness test showed that he accuracy of the autocoder was 92%, the accuracy of the hostile neural network was 96%, and in terms of sensitivity, the autocoder was 40% and the hostile neural network was 51%. In this paper, experiments have also been conducted to show how much performance changes due to differences in the optimization structure of potential variables. As a result, the level of 1% was improved in terms of sensitivity. These results suggest that it presented a new perspective on optimizing latent variable that were relatively insignificant.

• (The)Study of the Eastern Classic
• /
• no.66
• /
• pp.251-276
• /
• 2017

This article is an attempt to identify the difference between "Oneness between Heaven and Man(天人合一)" thought proposed by Zhoudunyi(周敦?) in 'Taijirushuo(太極圖說)' and "Cheon-In-Sim-Seong-Haap-Il(A diagram about Unifying Heaven, Human, and Mind-and-Heart)(天人心性合一)" thought presented by Kwonkun(權近) in 'Ip-Hak-Do-Seol(入學圖說)'. Oneness between Heaven and Man thought, which have developed into practical ideas centered on self-cultivation theory, have a philosophical depth and logical system by Zhoudunyi(周敦?) with 'Taijirushuo(太極圖說)' in order to summarize the metaphysical cosmology and the treatise of human nature in Dynasty Song. Zhoudunyi established the concept of Wuji(無極) corresponding to Heaven(天), and suggested unified cosmology of 'Wuji(無極)-Taiji(太極)-Yin and Yang(陰 陽)-Five elements(五行)-Human(人間)-Everything(萬物)'. His cosmology is the perception that heaven, the creator of all things, is the relationship of unity, which is connected by man as the creature and the Order(命)and Nature(性). Thus, when people restores one's nature and realized that it is a mandate from Heaven, he said that heavenly unity can be realized and become a Saint(聖人). The idea of Zhoudunyi(周敦?) comes to Zhu Xi(朱熹) about 120 years later and it is concluded to be converged the Neo-Confucianism with Lichi theory. Accordingly, Zhoudunyi(周敦?) is evaluated as the master of the Neo-Confucianism to posterity. KwonKun suggested the idea of "Cheon-In-Sim-Seong-Haap-Il(A diagram about Unifying Heaven, Human, and Mind-and-Heart)(天人心性合一)" in 'Ip-Hak-Do-Seol(入學圖說)' at the end of Goryeo Dynasty. KwonKun's "Cheon-In-Sim-Seong-Haap-Il" thought is logically coherent by adding an element of the doctrines of Chu-tzu to his teacher Yi Saek's "Chun-In-Mu-Gan" and arranging that the functions of my mind and Li(理) are all together. Whereas Zhuodunyi is concerned with the creation principle of the universe and all things, KwonKun mainly cares about Heaven(天), Human(人), and Heart (心), and Nature(性) in the view of psychology. In addition, he suggested that "Cheon-In-Sim-Seong-Haap-Il" can be achieved by self-cultivation centered of Gyeong-thought(敬思想). The idea of KwonKun has become a stepping stone to the development of the doctrines of Chu-tzu based on psychology. His ideas came to Yi Hwang about 150 year later, and are integrated into four clue-seven emotion (Sadan Chiljung四端七情) thesis and Gyeong-thought(敬思想). However, unlike the Zhoudunyi, KwonKun is not properly evaluated despite his academic accomplishments.

• Journal of Internet Computing and Services
• /
• v.21 no.6
• /
• pp.23-31
• /
• 2020

This study was carried out to generate various images of railroad surfaces with random defects as training data to be better at the detection of defects. Defects on the surface of railroads are caused by various factors such as friction between track binding devices and adjacent tracks and can cause accidents such as broken rails, so railroad maintenance for defects is necessary. Therefore, various researches on defect detection and inspection using image processing or machine learning on railway surface images have been conducted to automate railroad inspection and to reduce railroad maintenance costs. In general, the performance of the image processing analysis method and machine learning technology is affected by the quantity and quality of data. For this reason, some researches require specific devices or vehicles to acquire images of the track surface at regular intervals to obtain a database of various railway surface images. On the contrary, in this study, in order to reduce and improve the operating cost of image acquisition, we constructed the 'Defective Railroad Surface Regeneration Model' by applying the methods presented in the related studies of the Generative Adversarial Network (GAN). Thus, we aimed to detect defects on railroad surface even without a dedicated database. This constructed model is designed to learn to generate the railroad surface combining the different railroad surface textures and the original surface, considering the ground truth of the railroad defects. The generated images of the railroad surface were used as training data in defect detection network, which is based on Fully Convolutional Network (FCN). To validate its performance, we clustered and divided the railroad data into three subsets, one subset as original railroad texture images and the remaining two subsets as another railroad surface texture images. In the first experiment, we used only original texture images for training sets in the defect detection model. And in the second experiment, we trained the generated images that were generated by combining the original images with a few railroad textures of the other images. Each defect detection model was evaluated in terms of 'intersection of union(IoU)' and F1-score measures with ground truths. As a result, the scores increased by about 10~15% when the generated images were used, compared to the case that only the original images were used. This proves that it is possible to detect defects by using the existing data and a few different texture images, even for the railroad surface images in which dedicated training database is not constructed.