• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.1
• /
• pp.51-58
• /
• 2023

In this paper, we propose an organ segmentation technique for the automatic extraction of medical diagnostic indicators from X-ray images. In order to calculate diagnostic indicators of heart disease and spinal disease such as VHS(vertebral heart scale) and Cobb angle, it is necessary to accurately segment the thoracic spine, carina, and heart in a chest X-ray image. A deep neural network model in which the high-resolution representation of the image for each layer and the structure converted into a low-resolution feature map are connected in parallel was adopted. This structure enables the relative position information in the image to be effectively reflected in the segmentation process. It is shown that learning performance can be improved by combining the OCR module, in which pixel information and object information are mutually interacted in a multi-step process, and the channel attention module, which allows each channel of the network to be reflected as different weight values. In addition, a method of augmenting learning data is presented in order to provide robust performance against changes in the position, shape, and size of the subject in the X-ray image. The effectiveness of the proposed theory was evaluated through an experiment using 145 human chest X-ray images and 118 animal X-ray images.

• Korean Chemical Engineering Research
• /
• v.53 no.2
• /
• pp.236-242
• /
• 2015

We present two data-driven modeling methods, partial least square (PLS) and artificial neural network (ANN), to predict the major operating and performance variables of a polymer electrolyte membrane (PEM) fuel cell stack. PLS and ANN models were constructed using the experimental data obtained from the testing of a 30 kW-class PEM fuel cell stack, and then were compared with each other in terms of their prediction and computational performances. To reduce the complexity of the models, we combined a variables importance on PLS projection (VIP) as a variable selection method into the modeling procedure in which the predictor variables are selected from a set of input operation variables. The modeling results showed that the ANN models outperformed the PLS models in predicting the average cell voltage and cathode outlet temperature of the fuel cell stack. However, the PLS models also offered satisfactory prediction performances although they can only capture linear correlations between the predictor and output variables. Depending on the degree of modeling accuracy and speed, both ANN and PLS models can be employed for performance predictions, offline and online optimizations, controls, and fault diagnoses in the field of PEM fuel cell designs and operations.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.1
• /
• pp.163-169
• /
• 2020

Recently, research on automation and unmanned operation of machines in the industrial field has been conducted with the advent of AI, Big data, and the IoT, which are the core technologies of the Fourth Industrial Revolution. The machines for these automation processes are controlled based on the data collected from the sensors attached to them, and further, the processes are managed. Conventionally, the abnormalities of sensors are periodically checked and managed. However, due to various environmental factors and situations in the industrial field, there are cases where the inspection due to the failure is not missed or failures are not detected to prevent damage due to sensor failure. In addition, even if a failure occurs, it is not immediately detected, which worsens the process loss. Therefore, in order to prevent damage caused by such a sudden sensor failure, it is necessary to identify the failure of the sensor in an embedded system in real-time and to diagnose the failure and determine the type for a quick response. In this paper, a deep neural network-based fault diagnosis system is designed and implemented using Raspberry Pi to classify typical sensor fault types such as erratic fault, hard-over fault, spike fault, and stuck fault. In order to diagnose sensor failure, the network is constructed using Google's proposed Inverted residual block structure of MobilieNetV2. The proposed scheme reduces memory usage and improves the performance of the conventional CNN technique to classify sensor faults.

• The Korean Journal of Applied Statistics
• /
• v.37 no.2
• /
• pp.239-253
• /
• 2024

This research applies both point-wise and pair-wise learning strategies within the learning-to-rank (LTR) framework to predict horse race rankings in Seoul. Specifically, for point-wise learning, we employ a linear model and random forest. In contrast, for pair-wise learning, we utilize tools such as RankNet, and LambdaMART (XGBoost Ranker, LightGBM Ranker, and CatBoost Ranker). Furthermore, to enhance predictions, race records are standardized based on race distance, and we integrate various datasets, including race information, jockey information, horse training records, and trainer information. Our results empirically demonstrate that pair-wise learning approaches that can reflect the order information between items generally outperform point-wise learning approaches. Notably, CatBoost Ranker is the top performer. Through Shapley value analysis, we identified that the important variables for CatBoost Ranker include the performance of a horse, its previous race records, the count of its starting trainings, the total number of starting trainings, and the instances of disease diagnoses for the horse.

• Journal of Broadcast Engineering
• /
• v.24 no.4
• /
• pp.592-601
• /
• 2019

For coronary procedures, X-ray angiogram images are useful for diagnosing and assisting procedures. It is challenging to accurately segment a coronary artery using only a single segmentation model in 2D X-ray images due to a complex structure of three-dimensional coronary artery, especially from phenomenon of vessels being broken in the middle or end of coronary artery. In order to solve these problems, the initial segmentation is performed using an existing single model, and the candidate regions for the sophisticate correction is estimated based on the initial segment, and the local patch-based correction is performed in the candidate regions. Through this research, not only the broken coronary arteries are re-connected, but also the distal part of coronary artery that is very thin is additionally correctly found. Further, the performance can be much improved by combining the proposed correction method with any existing coronary artery segmentation method. In this paper, the U-net, a fully convolutional network was chosen as a segmentation method and the proposed correction method was combined with U-net to demonstrate a significant improvement in performance through X-ray images from several patients.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.10
• /
• pp.887-895
• /
• 2014

Drowsy driving is a large proportion of the total car accidents. For this reason, drowsiness detection and warning system for drivers has recently become a very important issue. Monitoring physiological signals provides the possibility of detecting features of drowsiness and fatigue of drivers. Many researches have been published that to measure electroencephalogram(EEG) signals is the effective way in order to be aware of fatigue and drowsiness of drivers. The aim of this study is to extract drowsiness-related features from a set of EEG signals and to classify the features into three states: alertness, transition, and drowsiness. This paper proposes a drowsiness detection system using errors-in-variables(EIV) for extraction of feature vectors and multilayer perceptron (MLP) for classification. The proposed method evaluates robustness for noise and compares to the previous one using linear predictive coding (LPC) combined with MLP. From evaluation results, we conclude that the proposed scheme outperforms the previous one in the low signal-to-noise ratio regime.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.3
• /
• pp.705-711
• /
• 2013

QRS detection of ECG is the most popular and easy way to detect cardiac-disease. But it is difficult to analyze the ECG signal because of various noise types. Also in the healthcare system that must continuously monitor people's situation, it is necessary to process ECG signal in realtime. In other words, the design of algorithm that exactly detects QRS wave using minimal computation and classifies PVC by analyzing the persons's physical condition and/or environment is needed. Thus, efficient QRS detection and PVC classification based on profiling method is presented in this paper. For this purpose, we detected QRS through the preprocessing method using morphological filter, adaptive threshold, and window. Also, we applied profiling method to classify each patient's normal cardiac behavior through hash function. The performance of R wave detection, normal beat and PVC classification is evaluated by using MIT-BIH arrhythmia database. The achieved scores indicate the average of 99.77% in R wave detection and the rate of 0.65% in normal beat classification error and 93.29% in PVC classification.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.17 no.2
• /
• pp.10-20
• /
• 2013

To evaluate the safety of a beam structure, strains are measured as an indicator of structural states. However, unless strain sensors are installed exactly on where maximum or other representative strains occur, the techniques by which rational assessment through measured strains is accomplished are required. Thus, this study suggests a process to estimate strain distribution on the steel beam from discrete strains measured by sensors. In the presented technique, the targeted beam is regarded to be subjected to unknown loads so that applicability is enhanced. Final strain distribution is given as form of a function after regression analysis. To verify the performance of estimation, a bending test for steel beam on which distributed and concentrated loads simultaneously act is conducted. From the comparison between estimated and directly measured strains in the test, the curve of strain distribution and the strain at arbitrary location could be predicted within maximum relative error 3.32% and maximum absolute error of $2.32{\mu}{\varepsilon}$, respectively. Thus reliable and practical monitoring is expected to apply effectively for the steel beam structure.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.23 no.7
• /
• pp.80-86
• /
• 2019

As computer performance improves, research using deep learning are being actively carried out in various fields. Recently, deep learning technology has been applying to the safety evaluation for structures. In particular, the internal blades of a turbine structure requires experienced experts and considerable time to detect surface damages because of the difficulty of separation of the blades from the structure and the dark environmental condition. This study proposes a Faster R-CNN deep learning model that can detect surface damages on the internal blades, which is one of the primary elements of the turbine structure. The deep learning model was trained using image data with dent and punch damages. The image data was also expanded using image filtering and image data generator techniques. As a result, the deep learning model showed 96.1% accuracy, 95.3% recall, and 96% precision. The value of the recall means that the proposed deep learning model could not detect the blade damages for 4.7%. The performance of the proposed damage detection system can be further improved by collecting and extending damage images in various environments, and finally it can be applicable for turbine engine maintenance.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.19 no.5
• /
• pp.92-103
• /
• 2015

Box-Wilson experimental design method, known as central composite design, is the design of any information-gathering exercises where variation is present. This method was devised to gather as much data as possible in spite of the low design cost. This method was employed to model the effect of mixing factors on several performances of 60 MPa high strength self compacting concrete and to numerically calculate the optimal mix proportion. The nonlinear relations between factors and responses of HSSCC were approximated in the form of second order polynomial equation. In order to characterize five performances like compressive strength, passing ability, segregation resistance, manufacturing cost and density depending on five factors like water-binder ratio, cement content, fine aggregate percentage, fly ash content and superplasticizer content, the experiments were made at the total 52 experimental points composed of 32 factorial points, 10 axial points and 10 center points. The study results showed that Box-Wilson experimental design was really effective in designing the experiments and analyzing the relation between factor and response.