• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.48 no.5
• /
• pp.99-106
• /
• 2011

In many applications, wireless sensor networks could be thought as data-centric networks, and the sensor nodes are densely distributed over a large sensor field. The sensor nodes are normally vulnerable in terms of security since they are very often deployed in a hostile environment and open space. In this paper, we propose a scheme for contents-based anomaly detection in wireless sensor networks. In this scheme we use the characteristics of sensor networks where several nodes surrounding an event point can simultaneously detect the phenomenon occurring and the contents detected from these sensors are limited to inside a certain range. The proposed scheme consists of several phases; training, testing and refining phases. Anomaly candidates detected by the distance-based anomaly detection scheme in the testing phase are sent to the refining phase. They are then compared in the sink node with previously collected data set to improve detection performance in the refining phase. Our simulation results suggest the effectiveness of the proposed scheme in this paper evidenced by the improvements of the detection rate and the false positive rate.

• IEIE Transactions on Smart Processing and Computing
• /
• v.4 no.4
• /
• pp.195-201
• /
• 2015

For median filtering (MF) detection in altered digital images, this paper presents a new feature vector that is formed from autoregressive (AR) coefficients via an AR model of the gradients between the neighboring row and column lines in an image. Subsequently, the defined 10-D feature vector is trained in a support vector machine (SVM) for MF detection among forged images. The MF classification is compared to the median filter residual (MFR) scheme that had the same 10-D feature vector. In the experiment, three kinds of test items are area under receiver operating characteristic (ROC) curve (AUC), classification ratio, and minimal average decision error. The performance is excellent for unaltered (ORI) or once-altered images, such as $3{\times}3$ average filtering (AVE3), QF=90 JPEG (JPG90), 90% down, and 110% up to scale (DN0.9 and Up1.1) images, versus $3{\times}3$ and $5{\times}5$ median filtering (MF3 and MF5, respectively) and MF3 and MF5 composite images (MF35). When the forged image was post-altered with AVE3, DN0.9, UP1.1 and JPG70 after MF3, MF5 and MF35, the performance of the proposed scheme is lower than the MFR scheme. In particular, the feature vector in this paper has a superior classification ratio compared to AVE3. However, in the measured performances with unaltered, once-altered and post-altered images versus MF3, MF5 and MF35, the resultant AUC by 'sensitivity' (TP: true positive rate) and '1-specificity' (FN: false negative rate) is achieved closer to 1. Thus, it is confirmed that the grade evaluation of the proposed scheme can be rated as 'Excellent (A)'.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.3
• /
• pp.531-539
• /
• 2019

This paper proposes the training features for malware family analysis and analyzes the multi-classification performance of ensemble models. We construct training data by extracting API and DLL information from malware executables and use Random Forest and XGBoost algorithms which are based on decision tree. API, API-DLL, and DLL-CM features for malware detection and family classification are proposed by analyzing frequently used API and DLL information from malware and converting high-dimensional features to low-dimensional features. The proposed feature selection method provides the advantages of data dimension reduction and fast learning. In performance comparison, the malware detection rate is 93.0% for Random Forest, the accuracy of malware family dataset is 92.0% for XGBoost, and the false positive rate of malware family dataset including benign is about 3.5% for Random Forest and XGBoost.

• Journal of Biomedical Engineering Research
• /
• v.44 no.3
• /
• pp.211-217
• /
• 2023

Recently, many studies have introduced artificial intelligence systems in the surgical process to reduce the incidence and mortality of complications in patients. Bleeding is a major cause of operative mortality and complications. However, there have been few studies conducted on detecting bleeding in surgical videos. To advance the development of deep learning models for detecting intraoperative hemorrhage, three models have been trained and compared; such as, YOLOv5, RetinaNet50, and RetinaNet101. We collected 1,016 bleeding images extracted from five surgical videos. The ground truths were labeled based on agreement from two specialists. To train and evaluate models, we divided the datasets into training data, validation data, and test data. For training, 812 images (80%) were selected from the dataset. Another 102 images (10%) were used for evaluation and the remaining 102 images (10%) were used as the evaluation data. The three main metrics used to evaluate performance are precision, recall, and false positive per image (FPPI). Based on the evaluation metrics, RetinaNet101 achieved the best detection results out of the three models (Precision rate of 0.99±0.01, Recall rate of 0.93±0.02, and FPPI of 0.01±0.01). The information on the bleeding detected in surgical videos can be quickly transmitted to the operating room, improving patient outcomes.

• Journal of Yeungnam Medical Science
• /
• v.16 no.2
• /
• pp.169-180
• /
• 1999

Background: Uterine cervical cancer is the most common malignant tumor of the women in Korea. This study was undertaken to evaluate the usefulness of the cervicography as a screening test of cervical cancer. Materials and Methods: Cervicography was taken from 482 women at department of obstetrics and gynecology, at Yeungnam University Hospital from March 1, 1998 to October 31, 1999. Of the 482 women, 172 women were exc1uded from the study for various reasons, and 310 women completed the study. Three-hundred and ten women had cervical cytology (Papanicolaou smear), cervicography and colposcopy, and punch biopsy was undertaken if any of the test result was abnormal. Results: The most common age group was 35-39, and 40-44, 45-49 in order and most common reason for having a screening test was regular check for cervical cancer. The mean duration from the last Pap smear was 17.1 months, and 64 women(20.4%) never had any prior screening tests. Of the 310 women, 254 women were categorized as normal or having benign disease such as cervicitis, erosion or metaplasia. Biopsy was taken from 56 patients and the results were 26 chronic cervicitis, 4 mild dysplasia, 6 moderate dysplasia, 2 severe dysplasia, 14 carcinoma in situ and 4 invasive carcinoma. The results of cytology and cervicography were well correlated(p<0.05). The sensitivity and specificity of cytology were 86.7% and 76.9%, respectively and the sensitivity and specificity of cervicography were 56.7% and 96.2%, respectively. False negative rate of cervicography(43.3%) was much higher than those of cytology(13. 3%) (p<0.05), but false positive rate of cervicography(3.8%) was much lower than that of cytology(23.1%) (p<0.05). Conclusion: It seems inappropriate to use cervicography as a single screening test for cervival cancer, but it may be an effective complementary test for cytology to lower the false negative rate of cytology.

• Childhood Kidney Diseases
• /
• v.11 no.1
• /
• pp.59-64
• /
• 2007

Purpose : Suprapubic aspiration(SPA) has been considered the "gold standard" for obtaining urine in non-toilet trained infants. Ultrasound(US)-guided SPA improves the success rate of the procedure and reduces the complications. However, many physicians perceive SPA as invasive and prefer the use of urethral catheterization (Ucat). We compared the success rate, complications and accuracy of US-guided SPA and Ucat. Methods : 121 infants who visited Ewha Womans University Mokdong Hospital with suspected urinary tract infection(UTI) were investigated. For the first study, the study infants were randomly assigned to either the US-guided SPA(n=32) or Ucat(n=32) groups. The success rate and complications of both procedures were compared. For the second study, US-guided SPA and Ucat were performed simultaneously(n=57). The accuracy of urethral catheterization was subsequently analyzed. The criteria for success was defined as the collection of more than 0.5 mL of urine. UTI was diagnosed by the presence of uropathogens over 105 colony-forming units (CFU)/mL. Results : The overall success rate of the US-guided SPA was 96.9%(71.9% in first attempts, 25.0% in second attempts) which was not significantly different compared to 96.9%(90.6% in first attempts, 6.3% in second attempts) in the Ucat(P>0.05) group. The aspirated urine volume was $7.4{\pm}3.7mL$ in the US-guided SPA group, which was not significantly different to $4.5{\pm}2.6mL$ in the Ucat(P>0.05) group. The accuracy of Ucat in comparison to the US guided SPA was low with sensitivity 59.5%, specificity 86.6%, false-positive rate 13.3% and false-negative rate 40.5%. Conclusion : US-guided SPA should be encouraged as the best method to collect the urine in non-toilet trained infants with UTI.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.17 no.2
• /
• pp.95-100
• /
• 2013

Purpose: Since 2008, hepatitis A patients was rapidly increasing. So, Most of the health checkup examinees were interested in whether hepatitis A antibody was a lot. thereby The number of tests was increasing. In recent years, Antibody test results in the range of cut-off values were increased. According to the cause analysis, most examinees had a hepatitis A vaccine. This study was conducted to classify hepatitis A antibody as natural antibody and antibody after vaccination and compared the titer for seroconversion rate based on cut-off values. Materials and Methods: For a month in August 2012, First, We surveyed 185 health examinees and classified 119 health examinees who had acquired natural antibody. Second, for employees who were inoculated against hepatitis at our hospital, We classified into 53 primary inoculators and 59 secondary inculators. when the standard of cut-off value was 1, The seroconversion rate was compared the titer divided by 0.90-1.10 (${\pm}$), 0.60-0,89 (1+), 0.30-0.59 (2+), 0.01-0.29 (3+) and we compared the titer for seroconversion rate by each manufacturer after vaccination. Results: When the standard of cut-off value was 1, the titer of 119 health examinees who had acquired natural antibody was 0.90-1.10 (${\pm}$): 0%, 0.60-0.89 (1+): 0%, 0.30-0.59 (2+): 4.2%, 0.01-0.29 (3+): 96% and the titer of <0.60 ($${\geq_-}2+$$) was 100%. The titer of 53 primary inoculators was 0.90-1.10 (${\pm}:59.1%$), 0.60-0.89 (1+): 18.1%, 0.30-0.59 (2+): 18.1%, 0.01-0.29 (3+): 4.6% and the seroconversion rate was 45.3%. The titer of $${\geq_-}0.60$$ ($${\leq_-}1+$$) was 77.3%. The titer of 59 secondary inoculators was 0.90-1.10 (${\pm}:1.9%$), 0.60-0.89 (1+): 15.4%, 0.30-0.59 (2+): 36.54%, 0.01-0.29 (3+): 46.2% and the seroconversion rate was 88.1%. The titer of <0.60 ($${\geq_-}2+$$) was 82.7%. When we compared the titer for seroconversion rate by each manufacturer after vaccination, the seroconversion rate of 53 primary inoculators was BNIBT: 20.8% (${\pm}:24.5%$), GB: 15.7% (${\pm}:7.8%$), RIAKEY: 94.3% (${\pm}:3.8%$), ROCHE: 83% (${\pm}:0%$), ABBOTT: 73.1% (${\pm}:5.8%$) and the seroconversion rate of 59 secondary inoculators was BNIBT : 86.4% (${\pm}:1.7%$), GB: 88.5% (${\pm}:1.9%$), RIAKEY: 100% (${\pm}:0%$), ROCHE: 98.3% (${\pm}:0%$), ABBOTT: 98.2% (${\pm}:0%$). Conclusion: The study show that the titer of natural immune antibodies is higher than the titer of vaccination and the titer of secondary inoculation is mainly higher than the titer of primary inoculation. Consequently, if we know the titer of hepatitis A antibodies, it will help to give resullt reports. And then, when we compared the titer and the seroconversion rate by each manufacturer, There was a very distinct difference. As the test subjects inoculate against hepatitis A (HAV), it is considered BNIBT, GB will occur false negative rate and RIAKEY, ROCHE, ABOTT will occur false positive rate.

• Journal of the Korean Society of Radiology
• /
• v.81 no.4
• /
• pp.886-898
• /
• 2020

Purpose The purpose of our study was to evaluate digital breast tomosynthesis as a breast cancer screening modality for women with gynecologic cancer. Materials and Methods This retrospective study included patients with underlying gynecologic malignancies who underwent screening digital breast tomosynthesis for breast cancer. The cancer detection rate, recall rate, sensitivity, specificity, and positive predictive value (PPV) were calculated. PPV1 was defined as the percentage of all positive screening exams that have a tissue diagnosis of cancer within a year. PPV2 was defined as the percentage of all diagnostic exams (and Breast Imaging Reporting and Data System category 4, 5 from screening setting) with a recommendation for tissue diagnosis that have cancer within a year. PPV3 was defined as the percentage of all known biopsies actually performed that resulted in a tissue diagnosis of cancer within the year. For each case of screen-detected cancer, we analyzed the age, type of underlying gynecologic malignancy, breast density, imaging features, final Breast Imaging Reporting and Data System assessment, histologic type, T and N stages, molecular subtype, and Ki-67 index. Results Among 508 patients, 7 with breast cancer were identified after a positive result. The cancer detection rate was 13.8 per 1000 screening exams, and the recall rate was 17.9%. The sensitivity was 100%, and the specificity was 83.2%. The false negative rate was 0 per 1000 exams. The PPV1, PPV2, and PPV3 were 7.7, 31.8, and 31.8, respectively. Conclusion Digital breast tomosynthesis may be a promising breast cancer screening modality for women with gynecologic cancer, based on the high cancer detection rate, high sensitivity, high PPV, and high detection rate of early-stage cancer observed in our study.

• Investigative Magnetic Resonance Imaging
• /
• v.9 no.2
• /
• pp.109-116
• /
• 2005

Purpose : To determine the usefulness of additional Mn-DPDP MRI for preoperative evaluation of the patients with colorectal cancers by comparison of dual-phase CT scan, Mn-DPDP enhanced MRI and combination of CT and MRI. Materials and Methods : Fifty-three colorectal cancer patients with 92 metastatic nodules underwent dualphase (arterial and portal) helical CT scan and Mn-DPDP MRI prior to surgery. The indication of MRI was presence or suspected of having metastatic lesions at CT scan and/or increased serum carcinoembryonic antigen (CEA) levels (10 ng/mL or more). The diagnosis was established by the combination of findings at surgery, intraoperative ultrasonography, and histopathologic examination. Two radiologists interpreted CT, MRI, and combination of CT-MRI at discrete sessions and evaluated each lesion for location, size, and intrinsic characteristics. The lesions were divided into three groups according to their diameter; 1cm<, 1-2 cm, and >2 cm. Diagnostic accuracy was evaluated using the alternative-free response receiver operating characteristic method. Detection and false positive rate were also evaluated. Results : In the lesions smaller than 1 cm, detection rate of combined CT-MRI was superior to CT or MRI alone (82%, p=0.036). The mean accuracy (Az values) of combined CT and MRI was significantly higher than that of CT in the lesions smaller than 2 cm (1 cm<, p=0.034; 1-2 cm, p=0.045). However, there was no significant difference between MRI and combined CT-MRI. The false positive rate of CT was higher than those of combined CT-MR in the lesions smaller than 1 cm (28%, p=0.023). Conclusion : Additional MRI using Mn-DPDP besides routine CT scan was helpful in differentiating the hepatic lesions (<2 cm) and could improve detection of the small hepatic metastases (<1 cm) from colorectal carcinoma.

• Sleep Medicine and Psychophysiology
• /
• v.1 no.1
• /
• pp.60-67
• /
• 1994

Objectives: The Obstructive sleep apnea syndrome is characterized by snoring, observed apnea during sleep and excessive daytime sleepiness. The overnight polysomnographic recording is used to investigate patients with possible obstructive sleep apnea syndrome. But the overnight polysomnographic recording is time consuming, expensive, and labor-intensive. Recently in diagnosis of obstructive sleep apnea syndrome, several simple screening tests have been suggested. Methods: This study was performed to assess the probability of application of clinical features and the degree of oxygen desaturation as a screening test for the obstructive sleep apnea syndrome. The sensitivity and specificity of the self-report of clinical features including snoring, observed apnea during sleep, excessive daytime sleepiness and insomnia were tested. And the degree of oxygen desaturation measured by oximetry in 42 subjects were compared with the overnight polysomnographic recording results. Results: In the prediction of apnea index more than 5, the sensitivity of observed apnea during sleep, snoring, excessive daytime sleepiness and insomnia were 96.8%, 93.5%, 38.7%, 25.8% and the specificity of those clinical features were 182%, 36.4%, 100%, 72.7%, respectively. In the prediction of apnea index more than 5, the sensitivity and specificity of the combination of more than three self-report clinical features were 54.8% and 90.9%. The degree of oxygen de saturation and maximal apnea duration in the group of apnea index more than 5 were significantly different from those in the group of apnea index below 5(P<0.001). And the apnea index was significantly correlated with the degree of oxygen desaturation and maximal apnea duration(P<0.001). Conclusion: These results suggest that application of clinical features alone as a screening test for the obstructive sleep apnea syndrome is inadequate because of it's high rate of false positive and false negative results. The degree of oxygen desaturation measured by oximetry is possibly applicable to screening test and follow up evaluation of treatment efficacy for the obstructive sleep apnea syndrome.