• Journal of Korea Multimedia Society
• /
• v.25 no.8
• /
• pp.1224-1232
• /
• 2022

Whole body bone scan is the most frequently performed nuclear medicine imaging to evaluate bone metastasis in cancer patients. We evaluated the performance of a VGG16-based transfer learning classifier for bone scan images in which metastatic bone lesion was present. A total of 1,000 bone scans in 1,000 cancer patients (500 patients with bone metastasis, 500 patients without bone metastasis) were evaluated. Bone scans were labeled with abnormal/normal for bone metastasis using medical reports and image review. Subsequently, gradient-weighted class activation maps (Grad-CAMs) were generated for explainable AI. The proposed model showed AUROC 0.96 and F1-Score 0.90, indicating that it outperforms to VGG16, ResNet50, Xception, DenseNet121 and InceptionV3. Grad-CAM visualized that the proposed model focuses on hot uptakes, which are indicating active bone lesions, for classification of whole body bone scan images with bone metastases.

• Journal of the Korean Society of Radiology
• /
• v.18 no.3
• /
• pp.187-201
• /
• 2024

Osteoporosis is a major health issue globally, often remaining undetected until a fracture occurs. To facilitate early detection, deep learning (DL) models were developed to classify osteoporosis using abdominal computed tomography (CT) scans. This study was conducted using retrospectively collected data from 3,012 contrast-enhanced abdominal CT scans. The DL models developed in this study were constructed for using image data, demographic/clinical information, and multi-modality data, respectively. Patients were categorized into the normal, osteopenia, and osteoporosis groups based on their T-scores, obtained from dual-energy X-ray absorptiometry, into normal, osteopenia, and osteoporosis groups. The models showed high accuracy and effectiveness, with the combined data model performing the best, achieving an area under the receiver operating characteristic curve of 0.94 and an accuracy of 0.80. The image-based model also performed well, while the demographic data model had lower accuracy and effectiveness. In addition, the DL model was interpreted by gradient-weighted class activation mapping (Grad-CAM) to highlight clinically relevant features in the images, revealing the femoral neck as a common site for fractures. The study shows that DL can accurately identify osteoporosis stages from clinical data, indicating the potential of abdominal CT scans in early osteoporosis detection and reducing fracture risks with prompt treatment.

• Journal of Radiation Protection and Research
• /
• v.13 no.1
• /
• pp.8-20
• /
• 1988

The annual individual and collective doses to the thyroids of four age-dependent groups due to the in-take of I-131 released from the Younggwang nuclear power plant NU-1 & 2, Korea, are estimated using the model presented in ICRP 29. Sensitivity and robustness of the model are analyzed. In case of 0.12% fuel defect during normal operation, the collective dose is founded to be 3.05${\times}10^{-3}$man-thyroid-Sv, which is higher than the value calculated by the GASPAR code, 2.3${\times}10^{-3}$man-thyroid-Sv. The maximal individual annual doses resulting from an acute release are higher than those calculated under the assumption of continuous release by $1.4{\sim}1.7$ times. The most important pathway to the infant is milk and, in contrast, that to child, teen and adult is ingestion of crops. The model used is the calculation appears to be influenced by the variables such as roubstness-index. The weighted committed dose equivalent obtained by the ICRP 29 model is slightly higher than that calculated by the three-compartment model.

• The korean journal of orthodontics
• /
• v.52 no.1
• /
• pp.3-19
• /
• 2022

Objective: The purpose of this study was to investigate the accuracy of one-step automated orthodontic diagnosis of skeletodental discrepancies using a convolutional neural network (CNN) and lateral cephalogram images with different qualities from nationwide multi-hospitals. Methods: Among 2,174 lateral cephalograms, 1,993 cephalograms from two hospitals were used for training and internal test sets and 181 cephalograms from eight other hospitals were used for an external test set. They were divided into three classification groups according to anteroposterior skeletal discrepancies (Class I, II, and III), vertical skeletal discrepancies (normodivergent, hypodivergent, and hyperdivergent patterns), and vertical dental discrepancies (normal overbite, deep bite, and open bite) as a gold standard. Pre-trained DenseNet-169 was used as a CNN classifier model. Diagnostic performance was evaluated by receiver operating characteristic (ROC) analysis, t-stochastic neighbor embedding (t-SNE), and gradient-weighted class activation mapping (Grad-CAM). Results: In the ROC analysis, the mean area under the curve and the mean accuracy of all classifications were high with both internal and external test sets (all, > 0.89 and > 0.80). In the t-SNE analysis, our model succeeded in creating good separation between three classification groups. Grad-CAM figures showed differences in the location and size of the focus areas between three classification groups in each diagnosis. Conclusions: Since the accuracy of our model was validated with both internal and external test sets, it shows the possible usefulness of a one-step automated orthodontic diagnosis tool using a CNN model. However, it still needs technical improvement in terms of classifying vertical dental discrepancies.

• Korean Journal of Veterinary Research
• /
• v.51 no.3
• /
• pp.227-231
• /
• 2011

This study was undertaken to assess the lateral ventricle, which was some portion of brain and related to congenital anomalies, from 1, 2, 4, and 8 months of age in healthy micropigs. They were induced general anesthesia and performed magnetic resonance imaging (MRI) with a 0.3 Tesla magnet. Each age group was evaluated by three subjects such as lateral ventricular volume, ventricular volume ratio and asymmetry. T1 weighted transverse images were acquired for calculation of lateral ventricular and corresponding brain parenchyma areas. The ratio of bilateral ventricle areas used to analyze the asymmetry. The mean ventricular volumes of each month were $676.74{\pm}25.58mm^3$ (1 month-old), $630.64{\pm}143.84mm^3$ (2 month-old), $992.12{\pm}106.03mm^3$ (4 month-old) and $1172.62{\pm}237.57mm^3$ (8 month-old), respectively. The ventricular volume ratio was the smallest at 2 month-old and re-increased from that age. The ratio was significantly different between 2 month-old and other age groups (p < 0.05). The value of bilateral area ratio showed within 1.5 in all experimental animals. Consequently the lateral ventricle showed a positive correlation with aging and symmetric shapes in both sides. The developmental pattern of the lateral ventricle provides basic data in micropigs as an experimental animal model for physiological and neurosurgical approach.

• Journal of Broadcast Engineering
• /
• v.13 no.3
• /
• pp.339-349
• /
• 2008

We present a novel method to estimate the light source direction and relight a face texture image of a single 3D model under arbitrary unknown illumination conditions. We create a virtual irradiance sphere to detect the light source direction from a given illuminated texture image using both normal vector mapping and weighted bilinear interpolation. We then induce a relighting equation with estimated ambient and diffuse coefficients. We provide the result of a series of experiments on light source estimation, relighting and face recognition to show the efficiency and accuracy of the proposed method in restoring the shading and shadows areas of a face texture image. Our approach for face relighting can be used for not only illuminant invariant face recognition applications but also reducing visual load and Improving visual performance in tasks using 3D displays.

• Journal of the Korea Society for Simulation
• /
• v.24 no.4
• /
• pp.59-68
• /
• 2015

In this paper, we proposed a methodology of vulnerability analysis for armored fighting vehicle based on modeling and simulation. The SES/MB framework serves hierarchical representation of the structure for a complex systems and is easy to conduct modeling for the armored fighting vehicle which consists of various components. When the armored fighting vehicle is hit by the shots from threat, the vulnerability of the armored fighting vehicle is decreased by damaged or penetrated level of armors and components. The penetration is determined by the result of comparing a penetration energy through penetration analysis equation and defence ability of armor and components. And the defence ability is determined in accordance with type and defined property of normal component and armor component, all components have a weighted values for the degree of importance. We developed a simulation program for verification proposed methodology. Thus, the program analyzes vulnerability for armored fighting vehicle SES/MB framework using importance.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.10 no.1
• /
• pp.69-79
• /
• 2008

Prediction of ground condition ahead of tunnel face might be the most important factor to prevent collapse during tunnel excavation. In this study, a non-destructive method to evaluate the phase velocity in model rock mass using wavelet transform of surface wave was proposed aiming at ground condition assessment ahead of tunnel face. Model tests using gypsum as a rocklike material composed of two layers were performed. A Piezoelectric actuator with frequencies ranging from 150 Hz to 5 kHz was selected as a harmonic source. The acceleration history was measured with two accelerometers. Wavelet transform analysis was used to obtain the dispersion curves from the measured data. The experimental results showed that the near-field effects can be neglected if the distance between two receivers is chosen to be three times the wavelength. A simple inversion method using weighted factor based on the normal distribution was proposed. The inversion results showed that the predicted phase velocity agreed reasonably well with the measured one when the wavelength influence factor was 0.2. The depth of propagation of surface wave was from 0.42 to 0.63 times the wavelength. The range of wavelength varying with phase velocity in dispersion curve matched well with that estimated by inversion technique.

• Journal of Korea Water Resources Association
• /
• v.49 no.8
• /
• pp.645-656
• /
• 2016

The study performed a regional flood frequency analysis and proposed a regression equation to estimate design floods corresponding to return periods for ungauged basins in Geum-river basin. Five preliminary tests were employed to investigate hydrological independence and homogeneity of streamflow data, i.e. the lag-one autocorrelation test, time homogeneity test, Grubbs-Beck outlier test, discordancy measure test ($D_i$), and regional homogeneity measure (H). The test results showed that streamflow data were time-independent, discordant and homogeneous within the basin. Using five probability distributions (generalized extreme value (GEV), three-parameter log-normal (LN-III), Pearson type 3 (P-III), generalized logistic (GLO), generalized Pareto (GPA)), comparative regional flood frequency analyses were carried out for the region. Based on the L-moment ratio diagram, average weighted distance (AWD) and goodness-of-fit statistics ($Z^{DIST}$), the GLO distribution was selected as the best fit model for Geum-river basin. Using the GLO, a regression equation was developed for estimating regional design floods, and validated by comparing the estimated and observed streamflows at the Ganggyeong station.

• Investigative Magnetic Resonance Imaging
• /
• v.7 no.2
• /
• pp.116-123
• /
• 2003

Purpose : The present study was undertaken to evaluate the usefulness of cerebral diffusion (DWI) and perfusion MR imaging (PWI) in rabbit models with hyperacute cerebral ischemic infarction. Materials and Methods : Experimental cerebral infarction were induced by direct injection of mixture of Histoacryl glue, lipiodol, and tungsten powder into the internal cerebral artery of 6 New-Zealand white rabbits, and they underwent conventional T1 and T2 weighted MR imaging, DWI, and PWI within 1 hour after the occlusion of internal cerebral artery. The PWI scan for each rabbit was obtained at the level of lateral ventricle and 1cm cranial to the basal ganglia. By postprocessing using special imaging software, perfusion images including cerebral blood volume (CBV), cerebral blood flow (CBF), and mean transit time (MTT) maps were obtained. The detection of infarcted lesion were evaluated on both perfusion maps and DWI. MTT difference time were measured in the perfusion defect lesion and symmetric contralateral normal cerebral hemisphere. Results : In all rabbits, there was no abnormal signal intensity on T2WI. But on DWI, abnormal high signal intensity, suggesting cerebral infarction, were detected in all rabbits. PWI (rCBV, CBF and MTT map) also showed perfusion defect in all rabbits. In four rabbits, the calculated square of perfusion defect in MTT map is larger than that of CBF map and in two rabbits, the calculated size of perfusion defect in MTT map and CBF map is same. Any rabbits do not show larger perfusion defect on CBF map than MTT map. In comparison between CBF map and DWI, 3 rabbits show larger square of lesion on CBF map than on DWI. The others shows same square of lesion on both technique. The size of lesion shown in 6 MTT map were larger than DWI. In three cases, the size of lesion shown in CBF map is equal to DWI. But these were smaller than MTT map. The calculated square of lesion in CBF map, equal to that of DWI and smaller than MTT map was three. And in one case, the calculated square of perfusion defect in MTT map was largest, and that of DWI was smallest. Conclusion : DWI and PWI may be useful in diagnosing hyperacute cerebral ischemic infarction and in e-valuating the cerebral hemodynamics in the rabbits.