컴퓨터 성능의 발전으로 빅데이터의 효율적인 사용이 가능해지면서, 심층 학습(deep learning)은 다양한 의료 분야에 활용할 수 있는 핵심적인 인공지능(artificial intelligence, AI) 기법으로 각광받고 있다. 이에 본 종설은 뇌종양 진단과 치료에 사용되는 자기공명영상(magnetic resonance imaging, MRI)의 심층 학습 기법을 소개하고자 하였다. 먼저 국내 AI의 의료 분야 도입의 동향을 분석하고, 이를 바탕으로 MRI를 활용한 뇌종양의 진단과 치료에 적용할 수 있는 심층 학습 기법과 그 결과들을 기술하였다. 뇌종양 진단과 치료 시, 심층 학습을 이용한 최근 사례는 영상 분류, 영상 품질 개선, 영상 분할로 나타났으며, 질병의 진단과 치료에 적용할 수 있는 객관적이고 높은 성능 수치를 나타내면서 그 유용성을 확인 할 수 있었다. 종합하자면, 심층 학습은 질병의 진단과 치료에 적용할 수 있는 유용한 지표이며, AI 역량을 지닌 의료진의 지도하에 점진적인 도입이 이뤄진다면 질병의 진단과 치료에 큰 도움을 주는 훌륭한 소프트웨어로 활용될 것으로 여겨진다. 본 종설이 심층 학습을 이해할 때 많은 도움이 되길 바라며, 향후 관련 연구를 수행할 때 가이드라인으로 활용될 것을 기대 한다.

In this work, we introduce a 100 kW class mobile plasma melting system designed for non-combustible radioactive wastes treatment. To ensure mobility, the designed system consists of two 24-ft commercial containers, each in charge of the plasma utilities and melting process. In the container for plasma utilities, a 100 kW class DC power supply is installed together with a chiller and gas supply system whereas the container for melting process has a transferred type arc melter as well as off-gas treatment system consisting of a heat exchanger, filtrations, scrubber and NOx removal system. As a heat source for a transferred type arc melter, we adopted a hollow electrode plasma torch with reverse polarity discharge structure. Detailed design for a 100 kW class mobile plasma melting system will be presented together with the main specifications of the components. In addition, the basic performance data of the melting system is also presented and discussed.

This study investigated the residue of tulathromycin (TLM) on the lung tissue of pigs intramuscularly injected with 2.5 mg/kg body weight as a single dose. Nine healthy cross swine were intramuscularly injected with the drug. Three treated animals were arbitrarily selected to be sacrificed at 7, 10 and 14 days after treatment. TLM residue concentrations in lung tissue were determined using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). The drug was extracted from lung samples using acetonitrile followed by clean-up with n-hexane. The analyte was separated on a Hypersil GOLDTM liquid chromatography column using 0.1% formic acid in deionized distilled water and acetonitrile. The correlation coefficient (R2) of the calibration curve was 0.9981, and the limits of detection and quantification were 10 and 50 μg/kg, respectively. Recoveries at three spiking levels ranged between 95.0-98.2%, and relative standard deviations were less than 3.95%. In TLM-treated group, the concentration of TLM on 7, 10 and 14 days post-treatment was 627.3, 496.9, 259.1 μg/kg, respectively. The developed method is sensitive and reliable for detection of TLM in porcine lung tissues. In addition, it can detect below the maximum residue limits in animal-derived food products destined for human consumption.

Clostridium perfringens (C. perfringens) may cause diarrhea and enterotoxemia in adult and young livestock, leading to problems in the production and management of farms. Four hundred fecal samples were collected from 25 goat farms located in Gyeongsangbuk-do Province in the Republic of Korea. Sixteen C. perfringens strains were isolates from fecal samples, and the isolates were identified as type A (n=11) and type D (n=5). Additionally, α- and ε-toxin genes were detected in 16 and 5 strains by PCR, respectively, and the enterotoxin gene was presented in 2 strains. The antibiotic susceptibility test was performed using the disk diffusion method and E-test method. In the disk diffusion method, ampicillin (n=16) and chloramphenicol (n=15) were highly susceptible to 16 C. perfringens isolates. In the E-test method, ampicillin, amoxicillin, amoxicillin/clavulanic acid and meropenem were susceptible to more than 14 of 16 C. perfringens isolates. This study indicates that administration of antibiotics such as ampicillin, amoxicillin/clavulanic acid and meropenem can prevent and treat C. perfringens infections in goats.

본 연구는 대전, 충청지역 초음파실 종사자를 대상으로 병원감염관리에 관한 설문 조사를 실시하여 분석 하였고, ATP를 이용하여 임상에서 사용되는 초음파 Probe의 오염도를 측정하였으며, 70% 알코올 이용하여 소독한 후 결과를 확인하였다. 기본적인 조사 대상자의 일반적인 특성 및 직종 별 대상자의 특성으로 나누어 학력, 직종, 초음파 전문자격증 보유 유무, 기관 내 감염 담당 부서 및 감염 교육의 여부를 분석하였다. 검사 후 젤 제거 여부 및 방법, Probe의 소독여부 및 변수 별 상관분석을 수행하여 초음파 Probe의 세균감염에 관한 인식 및 세척 및 소독에 대한 수행 정도를 분석하였다. 검사 후 Probe 세척에 관하여 수건을 가장 많이 사용하였으며, Gel 용기는 3개월 이상 교환하지 않는 것으로 나타났다. 70% 알코올 소독 후 ATP 오염도는 1055.4±944.2에서 133.5±93.2으로 감소하였고 통계적으로 유의한 것으로 분석되었다.(ρ<0.01) 병원 균종은 CNS, Gram positive bacillus, Micrococcus species 검출되었다. 이를 해결하기 위해 시중에서 손쉽게 구할 수 있는 70% 알코올을 이용하여 소독을 실시하였으며 소독 후 병원균종은 검출되지 않았다. 연구 결과 초음파실을 관리하는 의료인 또는 의료기사 등은 감염관리에 대한 주기적인 교육과 감염예방에 대한 노력이 필요하다고 볼 수 있으며, 70% 알코올은 병원균 사멸에 효과적인 것으로 나타났다. 따라서 의료기관은 적극적인 병원감염관리 교육을 실시하여 종사자들의 병원감염에 관한 인식을 개선하고, 환자 감염 예방에 기여해야 할 것이다. 또한 본 연구 결과를 적절히 활용한다면 초음파 Probe를 매개로 한 감염 예방에 도움이 될 것으로 사료된다.

방사선 방호의 목적 중 하나는 확률적 영향을 최소화 하는 것이다. PCXMC 2.0은 몬테카를로 시뮬레이션 기반의 프로그램으로 입사표면선량을 통해 유효선량과 암의 발병확률을 예측가능하게 해준다. 그렇기 때문에 선량계에 따른 입사표면선량 측정이 특히 중요하다. 본 연구는 반도체 선량계, 일반 선량계, 유리선량계를 통해 입사표면선량을 측정하고 그에 따른 결정 장기의 유효선량과 발병 확률을 비교분석 하는 것에 목적을 두었다. 실험방법은 두개부, 흉부, 복부의 선량계 별 입사표면선량을 측정하고 PCXMC 2.0을 통해 부위 별 결정 장기의 유효선량과 암의 발병 확률을 평가하였다. 그 결과 부위 별 입사표면선량은 동일한 조건임에도 일반 선량계, 반도체 선량계, 유리 선량계 순으로 차이가 났다. 이를 토대로 유효선량과 결정 장기의 암 발병 확률을 분석한 결과 또한 일반 선량계, 반도체 선량계, 유리 선량계 순으로 차이가 났다. 결론적으로 동일한 조건임에도 사용한 선량계에 따라 유효선량과 발병 위험도는 다르게 나타났음을 알 수 있었고, 본 연구를 통해 각각의 선량계에 따른 정확한 입사표면선량 모델을 제시하는 것이 중요하다는 것을 알 수 있었다.

In this study, groundwater flow was analyzed targeting Dae-jeong watershed, which exhibited the largest variations of groundwater levels at the identical elevation points among the 16 watersheds of Jeju Island. The issues of the methods applied in practice were identified and improvement plans were suggested. This groundwater-flow estimates derived by applying hydraulic conductivity values onto zones of equal topographic ground level were found to be quite different from actual measured groundwater flow. Conversely, groundwater-flow estimates that utilized hydraulic conductivity values applied onto groundwater-level equipotential lines indicated relatively lesser divergences from actual measured groundwater flow. The reliabilities of the two approaches were assessed for 60 randomly selected points on DEM (digital elevation model) maps, The method using hydraulic conductivity values applied onto groundwater-level contours turned out to be the more reliable approach for the Dae-jeong watershed in Jeju Island.

Jeju Island relies on subterranean water for over 98% of its water resources, and it is therefore necessary to continue to perform studies on drought due to climate changes. In this study, the representative standardized precipitation index (SPI) is classified by various criteria, and the spatial characteristics and applicability of drought in Jeju Island are evaluated from the results. As the result of calculating SPI of 4 weather stations (SPI 3, 6, 9, 12), SPI 12 was found to be relatively simple compared to SPI 6. Also, it was verified that the fluctuation of SPI was greater fot short-term data, and that long-term data was relatively more useful for judging extreme drought. Cluster analysis was performed using the K-means technique, with two variables extracted as the result of factor analysis, and the clustering was terminated with seven-time repeated calculations, and eventually two clusters were formed.

This study analyzed fluctuations of ground water level of ground water wells developed in Seongsan watershed of Jeju Island until 2013 using MODFLOW, a numerical analysis model. Ground water level shows greater fluctuations from increase of pump capacity compared to the number of ground water wells. The development of ground water at the top of watershed was found to have direct influence on ground water level. Ground water wells developed until 2013 were used to continue pumping for 50 days, and ground water level of coastal region was reduced below 50% compared to the standard water level. In addition, the range of fluctuation of water level was large in the east coast region, which represents the direction of flow of ground water.

Rainfall-runoff model of Jeju Oedo Stream was used to compute the optimal unit hydrograph by HEC-HMS model that reflecting on watershed characteristics. Each rainfall event was comparatively analyzed with the actual flow measurement using Clark, Snyder and SCS synthetic methods for derived unit hydrograph. Subsequently, the null hypothesis was established as p-value for peak flow and peak time of each unit hydrograph by one-way ANOVA(Analysis of variance) was larger than significance level of 0.05. There was no significant difference in peak flow and peak time between different methods of unit hydrograph. As a result of comparing error rate with actual flow measurement data, Clark synthetic unit graph best reflected in Oedo Stream as compared to other methods, and error rate of Clark unit hydrograph was 0.02~1.93% and error rate at peak time was 0∼2.74%.

As occurrence of gradually increasing extreme temperature events in Jeju Island, a hybrid downscaling technique that simultaneously applies by dynamical method and statistical method has implemented on design rainfall in order to reduce flood damages from severe storms and typhoons.As a result of computation, Case 1 shows a strong tendency to excessively compute rainfall, which is continuously increasing. While Case 2 showed similar trend as Case 1, low design rainfall has computed by rainfall in A1B scenario. Based on the design rainfall computation method mainly used in Preventive Disaster System through Pre-disaster Effect Examination System and Basic Plan for River of Jeju Island which are considering climatic change for selecting 50-year and 100-year frequencies. Case 3 selecting for Jeju rain gage station and Case 1 for Seogwipo rain gage station. The results were different for each rain gage station because of difference in rainfall characteristics according to recent climatic change, and the risk of currently known design rainfall can be increased in near future.

In the mountain streams in Jeju Island, strong turbulence and roughness usually made it nearly impossible to utilize most of intrusive instrumentation for streamflow discharge measurements. Instead, a non-intrusive fixed electro-magnetic wave surface velocimetry (fixed EWSV: Kalesto) became alternatively popular in many representative streams to measure stream discharge seamlessly. Currently, Kalesto has shown noteworthy performance with little loss in flood discharge measurements and also has successfully provided discharge for every minute. However, Kalesto has been operated to regard its measured one-point velocity as the representative mean velocity for the given cross-section. Therefore, it could be highly possible to potentially encompass discharge measurements errors. In this study, we analyzed the difference between such Kalesto discharge measurements and other alternative concurrent discharge measurements such as Acoustic Doppler Current Profiler (ADCP) and mobile EWSV which were able to measure velocity in multi-points in the cross-section. Consequently, Kalesto discharge deviated from ADCP discharge in amount of 48% for relatively low flow, and more than 20% difference for high flow compared with mobile EWSV discharge measurements. These results indicated that the one-point velocity measured by Kalesto should be used as a cross-sectional mean velocity, rather it should be accounted for as an index-velocity in conjunction with directly measured cross-sectional mean velocity by using more reliable instrumentations. After inducing Kalesto Discharge Correction Coefficient (KDCC) that actually means relationship between index velocity and cross-sectional mean velocity, the corrected discharge from Kalesto was significantly improved. Therefore, we found that index velocity method should be applied to obtain better accuracy of discharge measurement in case of Kalesto operation.