• Korean Journal of Radiology
• /
• v.21 no.7
• /
• pp.869-879
• /
• 2020

Objective: To evaluate the performance of a convolutional neural network (CNN) model that can automatically detect and classify rib fractures, and output structured reports from computed tomography (CT) images. Materials and Methods: This study included 1079 patients (median age, 55 years; men, 718) from three hospitals, between January 2011 and January 2019, who were divided into a monocentric training set (n = 876; median age, 55 years; men, 582), five multicenter/multiparameter validation sets (n = 173; median age, 59 years; men, 118) with different slice thicknesses and image pixels, and a normal control set (n = 30; median age, 53 years; men, 18). Three classifications (fresh, healing, and old fracture) combined with fracture location (corresponding CT layers) were detected automatically and delivered in a structured report. Precision, recall, and F1-score were selected as metrics to measure the optimum CNN model. Detection/diagnosis time, precision, and sensitivity were employed to compare the diagnostic efficiency of the structured report and that of experienced radiologists. Results: A total of 25054 annotations (fresh fracture, 10089; healing fracture, 10922; old fracture, 4043) were labelled for training (18584) and validation (6470). The detection efficiency was higher for fresh fractures and healing fractures than for old fractures (F1-scores, 0.849, 0.856, 0.770, respectively, p = 0.023 for each), and the robustness of the model was good in the five multicenter/multiparameter validation sets (all mean F1-scores > 0.8 except validation set 5 [512 x 512 pixels; F1-score = 0.757]). The precision of the five radiologists improved from 80.3% to 91.1%, and the sensitivity increased from 62.4% to 86.3% with artificial intelligence-assisted diagnosis. On average, the diagnosis time of the radiologists was reduced by 73.9 seconds. Conclusion: Our CNN model for automatic rib fracture detection could assist radiologists in improving diagnostic efficiency, reducing diagnosis time and radiologists' workload.

• IEIE Transactions on Smart Processing and Computing
• /
• v.1 no.1
• /
• pp.17-26
• /
• 2012

This paper presents an image processing-based validation method for unrecognizable numbers in severely distorted license plate images which have been degraded by various factors including low-resolution, low light-level, geometric distortion, and periodic noise. Existing vehicle license plate recognition (LPR) methods assume that most of the image degradation factors have been removed before performing the recognition of printed numbers and letters. If this is not the case, conventional LPR becomes impossible. The proposed method adopts a novel approach where a set of reference number images are intentionally degraded using the same factors estimated from the input image. After a series of image processing steps, including geometric transformation, super-resolution, and filtering, a comparison using cross-correlation between the intentionally degraded reference and the input images can provide a successful identification of the visually unrecognizable numbers. The proposed method makes it possible to validate numbers in a license plate image taken under low light-level conditions. In the experiment, using an extended set of test images that are unrecognizable to human vision, the proposed method provides a successful recognition rate of over 95%, whereas most existing LPR methods fail due to the severe distortion.

• Journal of the Society of Naval Architects of Korea
• /
• v.50 no.4
• /
• pp.232-239
• /
• 2013

This paper provides the structure of a Reynolds Averaged Navier-Stokes(RANS) based simulation method and its validation results for the ship motion problem. The motion information of the hull computed from the equations of motion is considered in the momentum equations as the relative fluid motions with respect to a non-inertial coordinates system. A finite volume method is used to solve the governing equations, while the free surface is captured by using a two-phase level-set method and the realizable k-${\varepsilon}$ model is used for turbulence closure. For the validation of the present numerical approach, the numerical results of the resistance and motion tests for DTMB 5415 at two ship speeds are compared against available experimental data.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.156-156
• /
• 2019

In this study, a gated recurrent unit (GRU) network is constructed based on a deep neural network (DNN) with the aim of restoring the missing daily flow data in river basins. Lai Chau hydrological station is located upstream of the Da river basin (Vietnam) is selected as the target station for this study. Input data of the model are data on observed daily flow for 24 years from 1961 to 1984 (before Hoa Binh dam was built) at 5 hydrological stations, in which 4 gauge stations in the basin downstream and restoring - target station (Lai Chau). The total available data is divided into sections for different purposes. The data set of 23 years (1961-1983) was employed for training and validation purposes, with corresponding rates of 80% for training and 20% for validation respectively. Another data set of one year (1984) was used for the testing purpose to objectively verify the performance and accuracy of the model. Though only a modest amount of input data is required and furthermore the Lai Chau hydrological station is located upstream of the Da River, the calculated results based on the suggested model are in satisfactory agreement with observed data, the Nash - Sutcliffe efficiency (NSE) is higher than 95%. The finding of this study illustrated the outstanding performance of the GRU network model in recovering the missing flow data at Lai Chau station. As a result, DNN models, as well as GRU network models, have great potential for application within the field of hydrology and hydraulics.

• Journal of The Geomorphological Association of Korea
• /
• v.26 no.2
• /
• pp.1-14
• /
• 2019

This study analyzed the relativity between block stream and talus distributions by employing a likelihood ratio approach. Possible distribution sites for each debris slope landform were extracted by applying a spatial integration model, in which we combined fuzzy set model, Bayesian predictive model, and logistic regression model. Moreover, to verify model performance, a success rate curve was prepared by cross-validation. The results showed that elevation, slope, curvature, topographic wetness index, geology, soil drainage, and soil depth were closely related to the debris slope landform sites. In addition, all spatial integration models displayed an accuracy of over 90%. The accuracy of the distribution potential area map of the block stream was highest in the logistic regression model (93.79%). Eventually, the accuracy of the distribution potential area map of the talus was also highest in the logistic regression model (97.02%). We expect that the present results will provide essential data and propose methodologies to improve the performance of efficient and systematic micro-landform studies. Moreover, our research will potentially help to enhance field research and topographic resource management.

• Genomics & Informatics
• /
• v.21 no.4
• /
• pp.45.1-45.11
• /
• 2023

Nonalcoholic fatty liver disease (NAFLD) is a common type of chronic liver disease, with severity levels ranging from nonalcoholic fatty liver to nonalcoholic steatohepatitis (NASH). The extent of liver fibrosis indicates the severity of NASH and the risk of liver cancer. However, the mechanism underlying NASH development, which is important for early screening and intervention, remains unclear. Weighted gene co-expression network analysis (WGCNA) is a useful method for identifying hub genes and screening specific targets for diseases. In this study, we utilized an mRNA dataset of the liver tissues of patients with NASH and conducted WGCNA for various stages of liver fibrosis. Subsequently, we employed two additional mRNA datasets for validation purposes. Gene set enrichment analysis (GSEA) was conducted to analyze gene function enrichment. Through WGCNA and subsequent analyses, complemented by validation using two additional datasets, we identified five genes (BICC1, C7, EFEMP1, LUM, and STMN2) as hub genes. GSEA analysis indicated that gene sets associated with liver metabolism and cholesterol homeostasis were uniformly downregulated. BICC1, C7, EFEMP1, LUM, and STMN2 were identified as hub genes of NASH, and were all related to liver metabolism, NAFLD, NASH, and related diseases. These hub genes might serve as potential targets for the early screening and treatment of NASH.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.49 no.3
• /
• pp.211-215
• /
• 2004

Near infrared reflectance spectroscopy (NIRS) is a rapid and accurate analytical method for determining the composition of agricultural products and feeds. This study was conducted to determine tocopherol and tocotrienol contents in rice bran by using NIRS system. Total 80 rice bran samples previously analyzed by HPLC were scanned by NIRS and over 60 samples were selected for calibration and validation equation. A calibration equation calculated by MPLS(modified partial least squares) regression technique was developed and coefficient of determination for tocopheyol and tocotyienol content were 0.975 and 0.984, respectively, in calibration sets. Each calibration equation was fitted to validation set that was performed with the remaining samples not included is the calibration set, which showed high positive correlation both in tocopherol and tocotrienol content file. This results demonstrate that the developed NIRS equation can be practically used as a rapid screening method for quantification of tocopherol and tocotrienol contents in rice bran.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1511-1511
• /
• 2001

Energy content, expressed as calories per gram, is an important part of the evaluation and marketing of foods in developed countries. Currently accepted methods of measurement of energy by U.S. food labeling legislation include measurement of gross calories by bomb calorimetry with an adjustment for undigested protein and by calculation using specific factors for the energy values of protein, carbohydrate less the amount of insoluble dietary fiber, and total fat. The ability of NIRS to predict the energy value of diverse, processed and unprocessed cereal food products was investigated. NIR spectra of cereal products were obtained with an NIR Systems monochromator and the wavelength range used for analysis was 1104-2494 nm. Gross energy of the foods was measured by oxygen bomb calorimetry (Parr Manual No. 120) and expressed as calories per gram (CPGI, range 4.05-5.49 cal/g). Energy value was adjusted for undigested protein (CPG2, range 3.99-5.38 cal/g) and undigested protein and insoluble dietary fiber (CPG3, range 2.42-5.35 cal/g). Using a multivariate analysis software package (ISI International, Inc.) partial least squares models were developed for the prediction of energy content. The standard error of cross validation and multiple coefficient of determination for CPGI using modified partial least squares regression (n=127) was 0.060 cal/g and 0.95, respectively, and the standard error of performance, coefficient of determination, bias and slope using an independent validation set (n=59) were 0.057 cal/g, 0.98, -0.027 cal/g and 1.05 respectively. The PLS loading for factor 1 (Pearson correlation coefficient 0.92) had significant absorption peaks correlated to C-H stretch groups in lipid at 1722/1764 nm and 2304/2346 nm and O-H groups in carbohydrate at 1434 and 2076 nm. Thus the model appeared to be predominantly influenced by lipid and carbohydrate. Models for CPG2 and CPG3 showed similar trends with standard errors of performance, using the independent validation set, of 0.058 and 0.088 cal/g, respectively, and coefficients of determination of 0.96. Thus NIRS provides a rapid and efficient method of predicting energy content of diverse cereal foods.

• Spatial Information Research
• /
• v.21 no.2
• /
• pp.45-54
• /
• 2013

Analytical and simulation error models have the ability to describe (or realize) error-corrupted versions of spatial data. But the different approaches for modeling positional errors require an internal validation that ascertains whether the analytical and simulation error models predict correct positional errors in a defined set of conditions. This paper presents stochastic simulation models of a point and a line segm ent to be validated w ith analytical error models, which are an error ellipse and an error band model, respectively. The simulation error models populate positional errors by the Monte Carlo simulation, according to an assumed error distribution prescribed by given parameters of a variance-covariance matrix. In the validation process, a set of positional errors by the simulation models is compared to a theoretical description by the analytical error models. Results show that the proposed simulation models realize positional uncertainties of the same spatial data according to a defined level of positional quality.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.10
• /
• pp.6069-6075
• /
• 2013

Background: At present, the diagnosis of colorectal cancer (CRC) requires a colorectal biopsy which is an invasive procedure. We undertook this pilot study to develop an alternative method and potential new biomarkers for diagnosis, and validated a set of well-integrated tools called ClinProt to investigate the serum peptidome in CRC patients. Methods: Fasting blood samples from 67 patients diagnosed with CRC by histological diagnosis, 55 patients diagnosed with colorectal adenoma by biopsy, and 65 healthy volunteers were collected. Division was into a model construction group and an external validation group randomly. The present work focused on serum proteomic analysis of model construction group by ClinProt Kit combined with mass spectrometry. This approach allowed construction of a peptide pattern able to differentiate the studied populations. An external validation group was used to verify the diagnostic capability of the peptidome pattern blindly. An immunoassay method was used to determine serum CEA of CRC and controls. Results: The results showed 59 differential peptide peaks in CRC, colorectal adenoma and health volunteers. A genetic algorithm was used to set up the classification models. Four of the identified peaks at m/z 797, 810, 4078 and 5343 were used to construct peptidome patterns, achieving an accuracy of 100% (> CEA, P<0.05). Furthermore, the peptidome patterns could differentiate the validation group with high accuracy close to 100%. Conclusions: Our results showed that proteomic analysis of serum with MALDI-TOF MS is a fast and reproducible approach, which may provide a novel approach to screening for CRC.