Joint Weakly and Semi-Supervised Deep Learning for Localization and Classification of Masses in Breast Ultrasound Images.

We propose a framework for localization and classification of masses in breast ultrasound images. We have experimentally found that training convolutional neural network-based mass detectors with large, weakly annotated datasets presents a non-trivial problem, while overfitting may occur with those trained with small, strongly annotated datasets. To overcome these problems, we use a weakly annotated dataset together with a smaller strongly annotated dataset in a hybrid manner. We propose a systematic weakly and semi-supervised training scenario with appropriate training loss selection. Experimental results show that the proposed method can successfully localize and classify masses with less annotation effort. The results trained with only 10 strongly annotated images along with weakly annotated images were comparable to results trained from 800 strongly annotated images, with the 95% confidence interval (CI) of difference -3%-5%, in terms of the correct localization (CorLoc) measure, which is the ratio of images with intersection over union with ground truth higher than 0.5. With the same number of strongly annotated images, additional weakly annotated images can be incorporated to give a 4.5% point increase in CorLoc, from 80% to 84.50% (with 95% CIs 76%-83.75% and 81%-88%). The effects of different algorithmic details and varied amount of data are presented through ablative analysis.

Directions for and prospects of the Environmental Health Study in Korean National Industrial Complexes (EHSNIC): A proposal for the third phase of the EHSNIC

The Environmental Health Study in the Korean National Industrial Complexes (EHSNIC) is a project that aims to monitor the exposure and health effects of environmental pollution among residents of national industrial complexes, as well as propose appropriate environmental health measures. Since its launch in 2003, this project has been initiated in eight national industrial complexes. Currently, it is necessary to review the accomplishments and limitations of the phases 1 and 2 of this project, and establish the direction of the upcoming the phase 3. Thus, the present study has developed principles and goals for the phase 3, considering the rationale and justification of the EHSNIC, and presented specific research contents accordingly. In the phase 3, it is important to improve the methods for exposure assessment and evaluation of health effects, in order to identify clearly the association between the pollutants released from industrial complexes and their health impacts, to develop and to reinforce communication strategies to promote participation of residents of communities near industrial complexes. Nonetheless, it is also important to maintain the basic goal of continuously monitoring the level of exposure to and health effects of environmental pollutants.

Directions for and prospects of the Environmental Health Study in Korean National Industrial Complexes (EHSNIC): A proposal for the third phase of the EHSNIC

The Environmental Health Study in the Korean National Industrial Complexes (EHSNIC) is a project that aims to monitor the exposure and health effects of environmental pollution among residents of national industrial complexes, as well as propose appropriate environmental health measures. Since its launch in 2003, this project has been initiated in eight national industrial complexes. Currently, it is necessary to review the accomplishments and limitations of the phases 1 and 2 of this project, and establish the direction of the upcoming the phase 3. Thus, the present study has developed principles and goals for the phase 3, considering the rationale and justification of the EHSNIC, and presented specific research contents accordingly. In the phase 3, it is important to improve the methods for exposure assessment and evaluation of health effects, in order to identify clearly the association between the pollutants released from industrial complexes and their health impacts, to develop and to reinforce communication strategies to promote participation of residents of communities near industrial complexes. Nonetheless, it is also important to maintain the basic goal of continuously monitoring the level of exposure to and health effects of environmental pollutants.

Longitudinal trends of blood lead levels before and after leaded gasoline regulation in Korea

The objective of this study was to verify a change in the longitudinal trend of blood lead levels for the Korean population, before and after the regulation of leaded gasoline— which occurred between 1987 and 1993 in Korea. A total of 77 reports on blood lead levels among general Korean population between 1981 and 2014 were selected, and the results were summarized to have the variables of year, number of subjects, the subjects’ range in age, gender, and blood lead concentrations (arithmetic mean). The annual average atmospheric lead levels for four major cities (i.e., Seoul, Busan, Daegu and Gwangju) were collected from the Air Pollution Monitoring Database from 1991, and pilot studies from 1985 to 1990 before the national air quality monitoring system was launched in 1991. Blood lead levels were visualized in a bubble plot in which the size of each bubble represented the sample size of each study, and the annual average concentrations in ambient air were depicted on line graphs. Blood lead levels in the Korean population tended to gradually increase from the early 1980s (approximately 15-20 μg/dL) until 1990-1992 (20-25 μg/dL). Blood lead levels then began to rapidly decrease until 2014 ( < 2 μg/dL). Similar patterns were observed for both adults (≥20 years) and younger children/adolescents. The same longitudinal trend was observed in annual average atmospheric lead concentration, which suggests a significant correlation between air lead concentration and blood lead concentration in the general population. In conclusion, the regulation of leaded gasoline has significantly contributed to the rapid change in blood lead concentrations. And, the regulation of other sources of lead exposure should be considered to further decrease blood lead levels in the Korean population.

Tracking by Sampling and IntegratingMultiple Trackers.

We propose the visual tracker sampler, a novel tracking algorithm that can work robustly in challenging scenarios, where several kinds of appearance and motion changes of an object can occur simultaneously. The proposed tracking algorithm accurately tracks a target by searching for appropriate trackers in each frame. Since the real-world tracking environment varies severely over time, the trackers should be adapted or newly constructed depending on the current situation, so that each specific tracker takes charge of a certain change in the object. To do this, our method obtains several samples of not only the states of the target but also the trackers themselves during the sampling process. The trackers are efficiently sampled using the Markov Chain Monte Carlo (MCMC) method from the predefined tracker space by proposing new appearance models, motion models, state representation types, and observation types, which are the important ingredients of visual trackers. All trackers are then integrated into one compound tracker through an Interacting MCMC (IMCMC) method, in which the trackers interactively communicate with one another while running in parallel. By exchanging information with others, each tracker further improves its performance, thus increasing overall tracking performance. Experimental results show that our method tracks the object accurately and reliably in realistic videos, where appearance and motion drastically change over time, and outperforms even state-of-the-art tracking methods.

Polymorphisms in genes involved in innate immunity and susceptibility to benzene-induced hematotoxicity

Benzene, a recognized hematotoxicant and carcinogen, can damage the human immune system. We studied the association between single nucleotide polymorphisms (SNPs) in genes involved in innate immunity and benzene hematotoxicity in a cross-sectional study of workers exposed to benzene (250 workers and 140 controls). A total of 1,236 tag SNPs in 149 gene regions of six pathways were included in the analysis. Six gene regions were significant for their association with white blood cell (WBC) counts (MBP, VCAM1, ALOX5, MPO, RAC2, and CRP) based on gene-region (P < 0.05) and SNP analyses (FDR < 0.05). VCAM1 rs3176867, ALOX5 rs7099684, and MPO rs2071409 were the three most significant SNPs. They showed similar effects on WBC subtypes, especially granulocytes, lymphocytes, and monocytes. A 3-SNP block in ALOXE3 (rs7215658, rs9892383, and rs3027208) showed a global association (omnibus P = 0.0008) with WBCs even though the three SNPs were not significant individually. Our study suggests that polymorphisms in innate immunity genes may play a role in benzene-induced hematotoxicity; however, independent replication is necessary.

No Association Between Functional Polymorphisms in COMT and MTHFR and Schizophrenia Risk in Korean Population / 한국역학회지

OBJECTIVES: Common genetic SNPs in two genes, encoding catechol-O-methyltransferase (COMT) and methylenetetrahydrofolate reductase (MTHFR), which are interconnected with COMT gene regulation, have been reported to contribute to schizophrenia risk. In this study, we evaluated the association between functional polymorphisms in COMT and MTHFR and schizophrenia risk with a case-control study in a Korean population. METHODS: We performed a case-control study by genotyping analysis using 360 cases and 348 controls in Korean subjects to determine the association between functional polymorphisms in COMT and MTHFR and schizophrenia risk. RESULTS: Four functional SNPs in COMT (Val158Met and rs165599) and MTHFR (C677T and A1298C) were genotyped by primer extension assay. None of the genotype distributions for the four SNPs was significantly different between cases and controls. Stratified analysis did not show any significant gender difference for any polymorphism. In addition, we found no evidence of a gene-gene interaction in the analysis of combined genotypes. CONCLUSION: Our results suggest no significant association between the selected functional polymorphisms of COMT or MTHFR in Korean schizophrenia subjects. However, further studies are required to confirm our findings in a larger number of subjects.

Identification and Application of Biomarkers in Molecular and Genomic Epidemiologic Research / 예방의학회지

Biomarkers are characteristic biological properties that can be detected and measured in a variety of biological matrices in the human body, including the blood and tissue, to give an indication of whether there is a threat of disease, if a disease already exists, or how such a disease may develop in an individual case. Along the continuum from exposure to clinical disease and progression, exposure, internal dose, biologically effective dose, early biological effect, altered structure and/or function, clinical disease, and disease progression can potentially be observed and quantified using biomarkers. While the traditional discovery of biomarkers has been a slow process, the advent of molecular and genomic medicine has resulted in explosive growth in the discovery of new biomarkers. In this review, issues in evaluating biomarkers will be discussed and the biomarkers of environmental exposure, early biologic effect, and susceptibility identified and validated in epidemiological studies will be summarized. The spectrum of genomic approaches currently used to identify and apply biomarkers and strategies to validate genomic biomarkers will also be discussed.

Cytochrome P450 1A1 (CYP1A1) polymorphisms and breast ancer risk in Korean women

Cytochrome P450 1A1 (CYP1A1) is involved in the 2-hydroxylation of estrogen, the hormone that plays a critical role in the etiology of breast carcinoma. We evaluated the associations between two CYP1A1 polymorphisms [MspI (rs4646903); Ile462Val (rs1048943)] and breast cancer in a multicenter case-control study of 513 breast cancer cases and 447 controls in Korea. Women carrying the T allele of the CYP1A1 MspI polymorphism were found to have a 1.72-fold (95% CI 1.11-2.68) greater risk of developing breast cancer. No association was found between any CYP1A1 Ile462Val polymorphism and breast cancer. Haplotype analysis of the two loci showed that the CA haplotype was associated with the lowest risk of breast cancer, and CA/CA diplotypes were associated with a lower risk of breast cancer [OR = 0.28 (0.13-0.61)] than others/others diplotypes. Moreover, this reduced risk was more pronounced among women with a lower body mass index (BMI) [OR = 0.18 (0.06-0.58)] or with a shorter lifetime exposure to estrogen [OR = 0.23 (0.07-0.81)]. The results obtained suggest that the CYP1A1 MspI polymorphisms could affect susceptibility to breast cancer.

Obesity and genetic polymorphism of ERCC2 and ERCC4 as modifiers of risk of breast cancer

To evaluate the relationship of genetic polymorphisms of ERCC2 and ERCC4 genes, both involved in nucleotide excision repair (NER), and the risk of breast cancer, a hospital-based case-control study was conducted in Korea. Histologically confirmed breast cancer cases (n=574) and controls (n=502) with no present or previous history of cancer were recruited from three teaching hospitals in Seoul during 1995-2001. Information on selected characteristics was collected by interviewed questionnaire. ERCC2 Asp312Asn (G>A) was genotyped by single-base extension assay and ERCC4 Ser835Ser (T>C) by dynamic allele-specific hybridization system. Although no significant association was observed between the genetic polymorphisms and the risk of breast cancer, women with both ERCC2 A allele- and ERCC4 C allele-containing genotypes showed a 2.6-fold risk (95% CI: 1.02-6.48) of breast cancer compared to women concurrently carrying the ERCC2 GG and ERCC4 TT genotypes. The breast cancer risk increased as the number of "at risk" genotypes increased with a borderline significance (P for trend = 0.07). Interactive effect was also observed between ERCC4 genotype and body mass idnex (BMI) for the breast cancer risk; the ERCC4 C allele containing genotypes posed a 1.7-fold (95% CI: 0.96-2.93) breast cancer risk in obese women (BMI>25 kg/m2) with a borderline significance. Our finding suggests that the combined effect of ERCC2 Asp312Asn and ERCC4 Ser835Ser genotypes might be associated with breast cancer risk in Korean women.

Genetic polymorphism of CYP17 and breast cancer risk in Korean women

CYP17 gene is involved in steroidogenesis and steroid metabolism. Epidemiologic results on the association between the CYP17 polymorphism and breast cancer risk have been inconsistent. We examined the association between the MspAI polymorphism at +27 relative to the start of transcription in the 5'-untranslated region of CYP17 gene and breast cancer risk in Korean women. Four hundred and sixty-two incident cases and 337 controls were recruited from three teaching hospitals in Seoul during 1994-2001. Polymorphism of the CYP17 gene was determined by a single base extension assay. Demographic and lifestyle characteristics were identified using structured questionnaire. Age-adjusted (aOR) and multivariate odds ratios (mOR) and 95% confidence intervals (CI) were estimated by unconditional logistic regression. The proportions of A1/A1, A1/A2 and A2/A2 genotypes among controls were 20.8%, 45.1% and 34.1%, respectively. Compared to the A1/A1 genotype, A1/A2 or A2/A2 genotype was not statistically significantly associated with overall breast cancer risk (i.e., mOR=1.01, 95% CI=0.69-1.47 and mOR=0.76, 95% CI=0.51-1.14, respectively). However, a significant association between CYP17 A2/A2 genotype and breast cancer was observed among women aged 50 years or less (mOR=0.58, 95% CI=0.34-0.99, P=0.04) and leaner women (body mass index < 22 kg/m2) (mOR=0.48, 95% CI=0.23-0.97, P=0.04). Our results suggest that genetic polymorphism in 5'-untranslated region of CYP17 might play a role in breast cancer development in Korean women among younger women aged less than 50 or leaner women with body mass index less than 22 kg/m2.

Current Status of Genomic Epidemiology Research / 예방의학회지

Genomic epidemiology is defined as "an evolving field of inquiring that uses the systematic application of epidemiologic methods and approaches in population-based studies of the impact of human genetic variation on health and disease (Khoury, 1998) ". Most human diseases are caused by the intricate interaction among environmental exposures and genetic susceptibility factors. Susceptibility genes involved in disease pathogenesis are categorized into two groups: high penetrance genes (i.e., BRAC1, RB, etc.) and low penetrance genes (i.e., GSTs, Cyps, XRCC1, ets.), and low penetrance susceptibility genes has the higher priority for epidemiological research due to high population attributable risk. In this paper, the summarized results of the association study between single nucleotide polymorphisms (SNPs) and breast cancer in Korea were introduced and the international trends of genomic epidemiology research were reviewed with an emphasis on internet-based case-control and cohort consortium.

Glutathione S-transferase P1 Genetic Polymorphisms and Breast Cancer Risk / Journal of the Korean Cancer Association, 대한암학회지

PURPOSE: To evaluate the potential association between the GSTP1 genotype and the development of breast cancer, a hospital based case-control study was conducted in South Korea. MATERIALS AND METGODS: The study population consisted of 171 histologically confirmed incidents of breast cancer cases, and 171 age-matched controls with no present, or previous, history of cancer. A PCR method was used for the genotyping analyses, and statistical evaluation was performed by an unconditional logistic regression model. RESULTS: No association was observed in the study subjects, or the premenopausal women group with GSTP1 Val allele. However, postmenopausal women with GSTP1 Val allele had a reduced risk of breast cancer (OR=0.3, 95% CI=0.1~0.7). When the data were stratified, by the known risk factors of breast cancer, a significant interaction was observed between the GSTP1 genotype and alcohol consumption (p for interaction = 0.01); women with GSTP1 Val allele, that drank regularly, had a 3.0-fold increased risk of breast cancer (95% CI=1.1~7.9), whereas women with GSTP1 Val allele, that never drink, had protective effects (OR=0.4, 95% CI=0.2~0.8). CONCLUSION: Our findings suggest that GSTP1 Ile105Val polymorphism influences the individual susceptibility to breast cancer, and that this effect may be modified by alcohol consumption.

Interaction between Genetic Polymorphisms of CYP2E1 & NAT1 and Smoking in Lung Cancer Development (Preliminary report) / 대한암학회지

PURPOSE: The interactive effects of genetic polymorphisms of cytochrome P4502E1 (CYP2E1) & N-acetyltransferase 1 (NAT1) and smoking on lung cancer development were evaluated in hospital based case-control study. MATERIALS AND METHODS: Male lung cancer patients (N= 157) and the male patients with no present or previous history of systemic illnesses who visited the urology department (N=138) were recruited (1998-1999). CYP2E1 & NAT1 genotypes were determined by PCR-RFLP method using RsaI and MboII digestion, respectively. RESULTS: CYP2E1 c2 or NAT1 *10 allele did not increased the risk of lung cancer. Heavy smokers (35