• Korean Journal of Radiology
• /
• v.23 no.4
• /
• pp.389-401
• /
• 2022

Objective: This study aimed to determine a factor for predicting suboptimal image quality of the hepatobiliary phase (HBP) of gadoxetic acid-enhanced MRI in patients with extrahepatic bile duct (EHD) cancer before MRI examination. Materials and Methods: We retrospectively evaluated 259 patients (mean age ± standard deviation: 68.0 ± 8.3 years; 162 male and 97 female) with EHD cancer who underwent gadoxetic acid-enhanced MRI between 2011 and 2017. Patients were divided into a primary analysis set (n = 184) and a validation set (n = 75) based on the diagnosis date of January 2014. Two reviewers assigned the functional liver imaging score (FLIS) to reflect the HBP image quality. The FLIS consists of the sum of three HBP features, each scored on a 0-2 scale: liver parenchymal enhancement, biliary excretion, and signal intensity of the portal vein. Patients were classified into low-FLIS (0-3) or high-FLIS (4-6) groups. Multivariable analysis was performed to determine a predictor of low FLIS using serum biochemical and imaging parameters of cholestasis severity. The optimal cutoff value for predicting low FLIS was obtained using receiver operating characteristic analysis, and validation was performed. Results: Of the 259 patients, 140 (54.0%) and 119 (46.0%) were classified into the low-FLIS and high-FLIS groups, respectively. In the primary analysis set, total bilirubin was an independent factor associated with low FLIS (adjusted odds ratio per 1-mg/dL increase, 1.62; 95% confidence interval [CI], 1.32-1.98). The optimal cutoff value of total bilirubin for predicting low FLIS was 2.1 mg/dL with a sensitivity of 95.1% (95% CI: 88.9-98.4) and a specificity of 89.0% (95% CI: 80.2-94.9). In the validation set, the total bilirubin cutoff showed a sensitivity of 92.1% (95% CI: 78.6-98.3) and a specificity of 83.8% (95% CI: 68.0-93.8). Conclusion: Serum total bilirubin before acquisition of gadoxetic acid-enhanced MRI may help predict suboptimal HBP image quality in patients with EHD cancer.

• Journal of Conservation Science
• /
• v.36 no.6
• /
• pp.511-518
• /
• 2020

There have been reports that the "jeongcho (定礎)" letters of the foundation stone at the historical site No. 280 of the "Main Building of the Bank of Korea in Seoul" were written by Prince Ito Hirobumi (伊藤博文), the first Resident-General of Japan in Korea. An on-site investigation by an advisory group consisting of three experts in calligraphy; revealed that the two characters of '定礎' inscribed on the foundation stone are the characteristics of Ito Hirobumi's handwriting, judging from the writing style and habits observed in the collections of the Central Library of Hamamatsu City, Japan. It was reported that his writing was confirmed by the experts, but no basis was provided. To provide more objective and quantitative supporting data, rather than qualitative judgment based on feeling, it is necessary to present the basis for judgment through quantitative image comparison results through image analysis. In this paper, using image analysis software, Ito Hirobumi's calligraphy writing and the inscribed characters of the foundation stone were compared and analyzed to confirm the contents of the press release. The character comparison process and character area measurement results are a good example showing that if objective judgment basis data are needed in a similar situation, an objective judgment basis can be prepared through quantification using image analysis.

• Current Industrial and Technological Trends in Aerospace
• /
• v.8 no.2
• /
• pp.98-104
• /
• 2010

The demand for Satellite Images is continuously increasing owing to the various applications of optical satellite images. However, the acquisition of optical images has a limitation due to problems of weather and day & night. because an optical satellite makes images with reflections of sunlight. Therefore, SAR Satellite, which uses electromagnetic waves to make an image, gives increased demand to various applications. It also makes increased interest. In this paper, a calibration and validation system, which is an essential element for high quality Radar images, is studied.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.4
• /
• pp.860-880
• /
• 2024

Breast cancer is a major health concern affecting women and men globally. Early detection and accurate classification of breast cancer are vital for effective treatment and survival of patients. This study addresses the challenge of accurately classifying breast tumors using machine learning classifiers such as MLP, AdaBoostM1, logit Boost, Bayes Net, and the J48 decision tree. The research uses a dataset available publicly on GitHub to assess the classifiers' performance and differentiate between the occurrence and non-occurrence of breast cancer. The study compares the 10-fold and 5-fold cross-validation effectiveness, showing that 10-fold cross-validation provides superior results. Also, it examines the impact of varying split percentages, with a 66% split yielding the best performance. This shows the importance of selecting appropriate validation techniques for machine learning-based breast tumor classification. The results also indicate that the J48 decision tree method is the most accurate classifier, providing valuable insights for developing predictive models for cancer diagnosis and advancing computational medical research.

• The Korean Journal of Food And Nutrition
• /
• v.37 no.4
• /
• pp.197-210
• /
• 2024

This study aims to develop a deep learning model to monitor rice serving amounts in institutional foodservice, enhancing personalized nutrition management. The goal is to identify the best convolutional neural network (CNN) for detecting rice quantities on serving trays, addressing balanced dietary intake challenges. Both a vanilla CNN and 12 pre-trained CNNs were tested, using features extracted from images of varying rice quantities on white trays. Configurations included optimizers, image generation, dropout, feature extraction, and fine-tuning, with top-1 validation accuracy as the evaluation metric. The vanilla CNN achieved 60% top-1 validation accuracy, while pre-trained CNNs significantly improved performance, reaching up to 90% accuracy. MobileNetV2, suitable for mobile devices, achieved a minimum 76% accuracy. These results suggest the model can effectively monitor rice servings, with potential for improvement through ongoing data collection and training. This development represents a significant advancement in personalized nutrition management, with high validation accuracy indicating its potential utility in dietary management. Continuous improvement based on expanding datasets promises enhanced precision and reliability, contributing to better health outcomes.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.15 no.4
• /
• pp.197-204
• /
• 2020

The ImageNet dataset is a large scale dataset and contains various natural scene images. In this paper, we propose a convolutional neural network (CNN)-based weighted ensemble technique for the ImageNet classification task. First, in order to fuse several models, our technique uses weights for each model, unlike the existing average-based ensemble technique. Then we propose an algorithm that automatically finds the coefficients used in later ensemble process. Our algorithm sequentially selects the model with the best performance of the validation set, and then obtains a weight that improves performance when combined with existing selected models. We applied the proposed algorithm to a total of 13 heterogeneous models, and as a result, 5 models were selected. These selected models were combined with weights, and we achieved 3.297% Top-5 error rate on the ImageNet test dataset.

• Journal of information and communication convergence engineering
• /
• v.16 no.3
• /
• pp.173-178
• /
• 2018

Accurate classification of cloud images is a challenging task. Almost all the existing methods rely on hand-crafted feature extraction. Their limitation is low discriminative power. In the recent years, deep learning with convolution neural networks (CNNs), which can auto extract features, has achieved promising results in many computer vision and image understanding fields. However, deep learning approaches usually need large datasets. This paper proposes a deep learning approach for classification of cloud image patches on small datasets. First, we design a suitable deep learning model for small datasets using a CNN, and then we apply data augmentation and dropout regularization techniques to increase the generalization of the model. The experiments for the proposed approach were performed on SWIMCAT small dataset with k-fold cross-validation. The experimental results demonstrated perfect classification accuracy for most classes on every fold, and confirmed both the high accuracy and the robustness of the proposed model.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.1418-1420
• /
• 2003

This presentation summarizes the development of the integration information search system for a Test-bed area located in Daejeon. It will be used for the validation of software components developed for the high resolution satellite image processing. The system development utilizes the Java programming language and implements the web browse capabilities to search, manage, and augment the satellite image data, the Ground Control Point(GCP) data, the spectral information on land cover types, the atmospheric data, and the topographical map.

• International Journal of Internet, Broadcasting and Communication
• /
• v.15 no.1
• /
• pp.269-274
• /
• 2023

In this paper, we proposed a new UNet 3+ model for medical image segmentation. The proposed ensemble(E) UNet 3+ model consists of UNet 3+s of varying depths into one unified architecture. UNet 3+s of varying depths have same encoder, but have their own decoders. They can bridge semantic gap between encoder and decoder nodes of UNet 3+. Deep supervision was used for learning on a total of 8 nodes of the E-UNet 3+ to improve performance. The proposed E-UNet 3+ model shows better segmentation results than those of the UNet 3+. As a result of the simulation, the E-UNet 3+ model using deep supervision was the best with loss function values of 0.8904 and 0.8562 for training and validation data. For the test data, the UNet 3+ model using deep supervision was the best with a value of 0.7406. Qualitative comparison of the simulation results shows the results of the proposed model are better than those of existing UNet 3+.

• Smart Structures and Systems
• /
• v.33 no.4
• /
• pp.301-311
• /
• 2024

This paper proposes a deep convolutional long short-term memory (ConvLSTM)-based crack growth prediction technique for predictive maintenance of structures. Since cracks are one of the critical damage types in a structure, their regular inspection has been mandatory for structural safety and serviceability. To effectively establish the structural maintenance plan using the inspection results, crack propagation or growth prediction is essential. However, conventional crack prediction techniques based on mathematical models are not typically suitable for tracking complex nonlinear crack propagation mechanism on civil structures under harsh environmental conditions. To address the technical issue, a field data-driven crack growth prediction technique using ConvLSTM is newly proposed in this study. The proposed technique consists of the four steps: (1) time-series crack image acquisition, (2) target image stabilization, (3) deep learning-based crack detection and quantification and (4) crack growth prediction. The performance of the proposed technique is experimentally validated using a concrete mock-up specimen by applying step-wise bending loads to generate crack growth. The validation test results reveal the prediction accuracy of 94% on average compared with the ground truth obtained by field measurement.