• Korean Journal of Radiology
• /
• v.23 no.4
• /
• pp.466-478
• /
• 2022

Objective: ¹⁸F-fluorodeoxyglucose (FDG) PET/CT is often used for detecting malignancy in patients with newly diagnosed hemophagocytic lymphohistiocytosis (HLH), with acceptable sensitivity but relatively low specificity. The aim of this study was to improve the diagnostic ability of ¹⁸F-FDG PET/CT in identifying malignancy in patients with HLH by combining ¹⁸F-FDG PET/CT and clinical parameters. Materials and Methods: Ninety-seven patients (age ≥ 14 years) with secondary HLH were retrospectively reviewed and divided into the derivation (n = 71) and validation (n = 26) cohorts according to admission time. In the derivation cohort, 22 patients had malignancy-associated HLH (M-HLH) and 49 patients had non-malignancy-associated HLH (NM-HLH). Data on pretreatment ¹⁸F-FDG PET/CT and laboratory results were collected. The variables were analyzed using the Mann-Whitney U test or Pearson's chi-square test, and a nomogram for predicting M-HLH was constructed using multivariable binary logistic regression. The predictors were also ranked using decision-tree analysis. The nomogram and decision tree were validated in the validation cohort (10 patients with M-HLH and 16 patients with NM-HLH). Results: The ratio of the maximal standardized uptake value (SUVmax) of the lymph nodes to that of the mediastinum, the ratio of the SUVmax of bone lesions or bone marrow to that of the mediastinum, and age were selected for constructing the model. The nomogram showed good performance in predicting M-HLH in the validation cohort, with an area under the receiver operating characteristic curve of 0.875 (95% confidence interval, 0.686-0.971). At an appropriate cutoff value, the sensitivity and specificity for identifying M-HLH were 90% (9/10) and 68.8% (11/16), respectively. The decision tree integrating the same variables showed 70% (7/10) sensitivity and 93.8% (15/16) specificity for identifying M-HLH. In comparison, visual analysis of ¹⁸F-FDG PET/CT images demonstrated 100% (10/10) sensitivity and 12.5% (2/16) specificity. Conclusion: ¹⁸F-FDG PET/CT may be a practical technique for identifying M-HLH. The model constructed using ¹⁸F-FDG PET/CT features and age was able to detect malignancy with better accuracy than visual analysis of ¹⁸F-FDG PET/CT images.

• Korean Journal of Radiology
• /
• v.23 no.3
• /
• pp.343-354
• /
• 2022

Objective: To develop and evaluate a deep learning-based artificial intelligence (AI) model for detecting skull fractures on plain radiographs in children. Materials and Methods: This retrospective multi-center study consisted of a development dataset acquired from two hospitals (n = 149 and 264) and an external test set (n = 95) from a third hospital. Datasets included children with head trauma who underwent both skull radiography and cranial computed tomography (CT). The development dataset was split into training, tuning, and internal test sets in a ratio of 7:1:2. The reference standard for skull fracture was cranial CT. Two radiology residents, a pediatric radiologist, and two emergency physicians participated in a two-session observer study on an external test set with and without AI assistance. We obtained the area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity along with their 95% confidence intervals (CIs). Results: The AI model showed an AUROC of 0.922 (95% CI, 0.842-0.969) in the internal test set and 0.870 (95% CI, 0.785-0.930) in the external test set. The model had a sensitivity of 81.1% (95% CI, 64.8%-92.0%) and specificity of 91.3% (95% CI, 79.2%-97.6%) for the internal test set and 78.9% (95% CI, 54.4%-93.9%) and 88.2% (95% CI, 78.7%-94.4%), respectively, for the external test set. With the model's assistance, significant AUROC improvement was observed in radiology residents (pooled results) and emergency physicians (pooled results) with the difference from reading without AI assistance of 0.094 (95% CI, 0.020-0.168; p = 0.012) and 0.069 (95% CI, 0.002-0.136; p = 0.043), respectively, but not in the pediatric radiologist with the difference of 0.008 (95% CI, -0.074-0.090; p = 0.850). Conclusion: A deep learning-based AI model improved the performance of inexperienced radiologists and emergency physicians in diagnosing pediatric skull fractures on plain radiographs.

• Korean Journal of Radiology
• /
• v.23 no.12
• /
• pp.1241-1250
• /
• 2022

Objective: To conduct a simulation study to determine whether artificial intelligence (AI)-aided mammography reading can reduce unnecessary recalls while maintaining cancer detection ability in women recalled after mammography screening. Materials and Methods: A retrospective reader study was performed by screening mammographies of 793 women (mean age ± standard deviation, 50 ± 9 years) recalled to obtain supplemental mammographic views regarding screening mammography-detected abnormalities between January 2016 and December 2019 at two screening centers. Initial screening mammography examinations were interpreted by three dedicated breast radiologists sequentially, case by case, with and without AI aid, in a single session. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and recall rate for breast cancer diagnosis were obtained and compared between the two reading modes. Results: Fifty-four mammograms with cancer (35 invasive cancers and 19 ductal carcinomas in situ) and 739 mammograms with benign or negative findings were included. The reader-averaged AUC improved after AI aid, from 0.79 (95% confidence interval [CI], 0.74-0.85) to 0.89 (95% CI, 0.85-0.94) (p < 0.001). The reader-averaged specificities before and after AI aid were 41.9% (95% CI, 39.3%-44.5%) and 53.9% (95% CI, 50.9%-56.9%), respectively (p < 0.001). The reader-averaged sensitivity was not statistically different between AI-unaided and AI-aided readings: 89.5% (95% CI, 83.1%-95.9%) vs. 92.6% (95% CI, 86.2%-99.0%) (p = 0.053), although the sensitivities of the least experienced radiologists before and after AI aid were 79.6% (43 of 54 [95% CI, 66.5%-89.4%]) and 90.7% (49 of 54 [95% CI, 79.7%-96.9%]), respectively (p = 0.031). With AI aid, the reader-averaged recall rate decreased by from 60.4% (95% CI, 57.8%-62.9%) to 49.5% (95% CI, 46.5%-52.4%) (p < 0.001). Conclusion: AI-aided reading reduced the number of recalls and improved the diagnostic performance in our simulation using women initially recalled for supplemental mammographic views after mammography screening.

• Korean Journal of Radiology
• /
• v.24 no.11
• /
• pp.1142-1150
• /
• 2023

Objective: To evaluate ^99mtechnetium-three polyethylene glycol spacers-arginine-glycine-aspartic acid (^99mTc-3PRGD₂) single-photon emission computed tomography (SPECT)/computed tomography (CT) imaging for diagnosing lymph node metastasis of primary malignant lung neoplasms. Materials and Methods: We prospectively enrolled 26 patients with primary malignant lung tumors who underwent ^99mTc-3PRGD₂ SPECT/CT and ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET)/CT imaging. Both imaging methods were analyzed in qualitative (visual dichotomous and 5-point grades for lymph nodes and lung tumors, respectively) and semiquantitative (maximum tissue-to-background radioactive count) manners for the lymph nodes and lung tumors. The performance of the differentiation of lymph nodes with and without metastasis was determined at the per-lymph node station and per-patient levels using histopathological results as the reference standard. Results: Total 42 stations had metastatic lymph nodes and 136 stations had benign lymph nodes. The differences between metastatic and benign lymph nodes in the visual qualitative and semiquantitative analyses of ^99mTc-3PRGD₂ SPECT/CT and ¹⁸F-FDG PET/CT were statistically significant (all P < 0.001). The area under the receiver operating characteristic curve (AUC) in the semi-quantitative analysis of ^99mTc-3PRGD₂ SPECT/CT was 0.908 (95% confidence interval [CI], 0.851-0.966), and the sensitivity, specificity, positive predictive value, and negative predictive value were 0.86 (36/42), 0.88 (120/136), 0.69 (36/52), and 0.95 (120/126), respectively. Among the 26 patients (including two patients each with two lung tumors), 15 had pathologically confirmed lymph node metastasis. The difference between primary lung lesions in patients with and without lymph node metastasis was statistically significant only in the semi-quantitative analysis of ^99mTc-3PRGD₂ SPECT/CT (P = 0.007), with an AUC of 0.807 (95% CI, 0.641-0.974). Conclusion: ^99mTc-3PRGD₂ SPECT/CT imaging may notably perform in the direct diagnosis of lymph node metastasis of primary malignant lung tumors and indirectly predict the presence of lymph node metastasis through uptake in the primary lesions.

• Korean Journal of Radiology
• /
• v.20 no.5
• /
• pp.759-772
• /
• 2019

Objective: To investigate the value of ultrasound (US) microflow assessment in distinguishing malignant from benign solid breast masses as well as the association between US parameters and histologic microvessel density (MVD). Materials and Methods: Ninety-eight breast masses (57 benign and 41 malignant) were examined using Superb Microvascular Imaging (SMI) and contrast-enhanced US (CEUS) before biopsy. Two radiologists evaluated the quantitative and qualitative vascular parameters on SMI (vascular index, morphology, distribution, and penetration) and CEUS (time-intensity curve analysis and enhancement characteristics). US parameters were compared between benign and malignant masses and the diagnostic performance was compared between SMI and CEUS. Subgroup analysis was performed according to lesion size. The effect of vascular parameters on downgrading Breast Imaging Reporting and Data System (BI-RADS) category 4A masses was evaluated. The association between histologic MVD and US parameters was analyzed. Results: Malignant masses were associated with a higher vascular index (15.1 ± 7.3 vs. 5.9 ± 5.6), complex vessel morphology (82.9% vs. 42.1%), central vascularity (95.1% vs. 59.6%), penetrating vessels (80.5% vs. 31.6%) on SMI (all, p < 0.001), as well as higher peak intensity (37.1 ± 25.7 vs. 17.0 ± 15.8, p < 0.001), slope (10.6 ± 11.2 vs. 3.9 ± 4.2, p = 0.001), area (1035.7 ± 726.9 vs. 458.2 ± 410.2, p < 0.001), hyperenhancement (95.1% vs. 70.2%, p = 0.005), centripetal enhancement (70.7% vs. 45.6%, p = 0.023), penetrating vessels (65.9% vs. 22.8%, p < 0.001), and perfusion defects (31.7% vs. 3.5%, p < 0.001) on CEUS (p ≤ 0.023). The areas under the receiver operating characteristic curve (AUCs) of SMI and CEUS were 0.853 and 0.841, respectively (p = 0.803). In 19 masses measuring < 10 mm, central vascularity on SMI was associated with malignancy (100% vs. 38.5%, p = 0.018). Considering all benign SMI parameters on the BI-RADS assessment, unnecessary biopsies could be avoided in 12 category 4A masses with improved AUCs (0.500 vs. 0.605, p < 0.001). US vascular parameters associated with malignancy showed higher MVD (p ≤ 0.016). MVD was higher in malignant masses than in benign masses, and malignant masses negative for estrogen receptor or positive for Ki67 had higher MVD (p < 0.05). Conclusion: US microflow assessment using SMI and CEUS is valuable in distinguishing malignant from benign solid breast masses, and US vascular parameters are associated with histologic MVD.

• Annals of Occupational and Environmental Medicine
• /
• v.35
• /
• pp.24.1-24.15
• /
• 2023

Background: The construction workers are vulnerable to fatigue due to high physical workload. This study aimed to investigate the relationship between overwork and heart rate in construction workers and propose a scheme to prevent overwork in advance. Methods: We measured the heart rates of construction workers at a construction site of a residential and commercial complex in Seoul from August to October 2021 and develop an index that monitors overwork in real-time. A total of 66 Korean workers participated in the study, wearing real-time heart rate monitoring equipment. The relative heart rate (RHR) was calculated using the minimum and maximum heart rates, and the maximum acceptable working time (MAWT) was estimated using RHR to calculate the workload. The overwork index (OI) was defined as the cumulative workload evaluated with the MAWT. An appropriate scenario line (PSL) was set as an index that can be compared to the OI to evaluate the degree of overwork in real-time. The excess overwork index (EOI) was evaluated in real-time during work performance using the difference between the OI and the PSL. The EOI value was used to perform receiver operating characteristic (ROC) curve analysis to find the optimal cut-off value for classification of overwork state. Results: Of the 60 participants analyzed, 28 (46.7%) were classified as the overwork group based on their RHR. ROC curve analysis showed that the EOI was a good predictor of overwork, with an area under the curve of 0.824. The optimal cut-off values ranged from 21.8% to 24.0% depending on the method used to determine the cut-off point. Conclusion: The EOI showed promising results as a predictive tool to assess overwork in real-time using heart rate monitoring and calculation through MAWT. Further research is needed to assess physical workload accurately and determine cut-off values across industries.

• Korean Journal of Radiology
• /
• v.20 no.5
• /
• pp.791-800
• /
• 2019

Objective: To compare various models of diffusion-weighted imaging including monoexponential apparent diffusion coefficient (ADC), biexponential (fast diffusion coefficient [D_f], slow diffusion coefficient [D_s], and fraction of fast diffusion), stretched-exponential (distributed diffusion coefficient and anomalous exponent term [α]), and kurtosis (mean diffusivity and mean kurtosis [MK]) models in the differentiation of renal solid masses. Materials and Methods: A total of 81 patients (56 men and 25 women; mean age, 57 years; age range, 30-69 years) with 18 benign and 63 malignant lesions were imaged using 3T diffusion-weighted MRI. Diffusion model selection was investigated in each lesion using the Akaike information criteria. Mann-Whitney U test and receiver operating characteristic (ROC) analysis were used for statistical evaluations. Results: Goodness-of-fit analysis showed that the stretched-exponential model had the highest voxel percentages in benign and malignant lesions (90.7% and 51.4%, respectively). ADC, D_s, and MK showed significant differences between benign and malignant lesions (p < 0.05) and between low- and high-grade clear cell renal cell carcinoma (ccRCC) (p < 0.05). α was significantly lower in the benign group than in the malignant group (p < 0.05). All diffusion measures showed significant differences between ccRCC and non-ccRCC (p < 0.05) except D_f and α (p = 0.143 and 0.112, respectively). α showed the highest diagnostic accuracy in differentiating benign and malignant lesions with an area under the ROC curve of 0.923, but none of the parameters from these advanced models revealed significantly better performance over ADC in discriminating subtypes or grades of renal cell carcinoma (RCC) (p > 0.05). Conclusion: Compared with conventional diffusion parameters, α may provide additional information for differentiating benign and malignant renal masses, while ADC remains the most valuable parameter for differentiation of RCC subtypes and for ccRCC grading.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.3
• /
• pp.833-838
• /
• 2024

The urban railway sleeper floating track(STEDEF) is a structure that structurally separates the sleepers and the concrete bed using sleeper boots and resilience pads to reduce vibration transmitted to the concrete bed. Recently, the resilience pads of sleeper floating tracks that have been in use for more than 20 years are deteriorating. Accordingly, in order to evaluate the performance of the resilience pad, a static spring stiffness test is being performed after extracting the resilience pad. This evaluation technique is performed after replacing the resilience pad in use. However, the track natural frequency can change depending on the resilience pad spring stiffness and the uplift and subsidence of the concrete bed. In this study, modal testing technique was used to evaluate the track natural frequency. For this purpose, the sleeper boots material, resilience pad spring stiffness, and track natural frequency according to concrete bed uplift and subsidence were measured using modal tests at a laboratory scale. It was analyzed that the natural frequency of the sleeper floating track was directly affected by changes in the spring stiffness of the resilience pad. In addition, the change in natural frequency due to the uplift and subsidence of the concrete bed was also found to be large. Therefore, it is believed that the modal test technique presented in this study can be used to evaluate the resilience pad deterioration and voided sleepers.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.79-104
• /
• 2020

Recently, as deep learning has attracted attention, the use of deep learning is being considered as a method for solving problems in various fields. In particular, deep learning is known to have excellent performance when applied to applying unstructured data such as text, sound and images, and many studies have proven its effectiveness. Owing to the remarkable development of text and image deep learning technology, interests in image captioning technology and its application is rapidly increasing. Image captioning is a technique that automatically generates relevant captions for a given image by handling both image comprehension and text generation simultaneously. In spite of the high entry barrier of image captioning that analysts should be able to process both image and text data, image captioning has established itself as one of the key fields in the A.I. research owing to its various applicability. In addition, many researches have been conducted to improve the performance of image captioning in various aspects. Recent researches attempt to create advanced captions that can not only describe an image accurately, but also convey the information contained in the image more sophisticatedly. Despite many recent efforts to improve the performance of image captioning, it is difficult to find any researches to interpret images from the perspective of domain experts in each field not from the perspective of the general public. Even for the same image, the part of interests may differ according to the professional field of the person who has encountered the image. Moreover, the way of interpreting and expressing the image also differs according to the level of expertise. The public tends to recognize the image from a holistic and general perspective, that is, from the perspective of identifying the image's constituent objects and their relationships. On the contrary, the domain experts tend to recognize the image by focusing on some specific elements necessary to interpret the given image based on their expertise. It implies that meaningful parts of an image are mutually different depending on viewers' perspective even for the same image. So, image captioning needs to implement this phenomenon. Therefore, in this study, we propose a method to generate captions specialized in each domain for the image by utilizing the expertise of experts in the corresponding domain. Specifically, after performing pre-training on a large amount of general data, the expertise in the field is transplanted through transfer-learning with a small amount of expertise data. However, simple adaption of transfer learning using expertise data may invoke another type of problems. Simultaneous learning with captions of various characteristics may invoke so-called 'inter-observation interference' problem, which make it difficult to perform pure learning of each characteristic point of view. For learning with vast amount of data, most of this interference is self-purified and has little impact on learning results. On the contrary, in the case of fine-tuning where learning is performed on a small amount of data, the impact of such interference on learning can be relatively large. To solve this problem, therefore, we propose a novel 'Character-Independent Transfer-learning' that performs transfer learning independently for each character. In order to confirm the feasibility of the proposed methodology, we performed experiments utilizing the results of pre-training on MSCOCO dataset which is comprised of 120,000 images and about 600,000 general captions. Additionally, according to the advice of an art therapist, about 300 pairs of 'image / expertise captions' were created, and the data was used for the experiments of expertise transplantation. As a result of the experiment, it was confirmed that the caption generated according to the proposed methodology generates captions from the perspective of implanted expertise whereas the caption generated through learning on general data contains a number of contents irrelevant to expertise interpretation. In this paper, we propose a novel approach of specialized image interpretation. To achieve this goal, we present a method to use transfer learning and generate captions specialized in the specific domain. In the future, by applying the proposed methodology to expertise transplant in various fields, we expected that many researches will be actively conducted to solve the problem of lack of expertise data and to improve performance of image captioning.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.57-79
• /
• 2016

It is a rising trend that the number of users using one of the social media channels, the Social Network Service, so called the SNS, is getting increased. As per to this social trend, more companies have interest in this networking platform and start to invest their funds in it. It has received much attention as a tool spreading and expanding the message that a company wants to deliver to its customers and has been recognized as an important channel in terms of the relationship marketing with them. The environment of media that is radically changing these days makes possible for companies to approach their customers in various ways. Particularly, the social network service, which has been developed rapidly, provides the environment that customers can freely talk about products. For companies, it also works as a channel that gives customized information to customers. To succeed in the online environment, companies need to not only build the relationship between companies and customers but focus on the relationship between customers as well. In response to the online environment with the continuous development of technology, companies have tirelessly made the novel marketing strategy. Especially, as the one-to-one marketing to customers become available, it is more important for companies to maintain the relationship marketing with their customers. Among many SNS, Facebook, which many companies use as a communication channel, provides a fan page service for each company that supports its business. Facebook fan page is the platform that the event, information and announcement can be shared with customers using texts, videos, and pictures. Companies open their own fan pages in order to inform their companies and businesses. Such page functions as the websites of companies and has a characteristic of their brand communities such as blogs as well. As Facebook has become the major communication medium with customers, companies recognize its importance as the effective marketing channel, but they still need to investigate their business performances by using Facebook. Although there are infinite potentials in Facebook fan page that even has a function as a community between users, which other platforms do not, it is incomplete to regard companies' Facebook fan pages as communities and analyze them. In this study, it explores the relationship among customers through the network of the Facebook fan page users. The previous studies on a company's Facebook fan page were focused on finding out the effective operational direction by analyzing the use state of the company. However, in this study, it draws out the structural variable of the network, which customer committment can be measured by applying the social network analysis methodology and investigates the influence of the structural characteristics of network on the business performance of companies in an empirical way. Through each company's Facebook fan page, the network of users who engaged in the communication with each company is exploited and it is the one-mode undirected binary network that respectively regards users and the relationship of them in terms of their marketing activities as the node and link. In this network, it draws out the structural variable of network that can explain the customer commitment, who pressed "like," made comments and shared the Facebook marketing message, of each company by calculating density, global clustering coefficient, mean geodesic distance, diameter. By exploiting companies' historical performance such as net income and Tobin's Q indicator as the result variables, this study investigates influence on companies' business performances. For this purpose, it collects the network data on the subjects of 54 companies among KOSPI-listed companies, which have posted more than 100 articles on their Facebook fan pages during the data collection period. Then it draws out the network indicator of each company. The indicator related to companies' performances is calculated, based on the posted value on DART website of the Financial Supervisory Service. From the academic perspective, this study suggests a new approach through the social network analysis methodology to researchers who attempt to study the business-purpose utilization of the social media channel. From the practical perspective, this study proposes the more substantive marketing performance measurements to companies performing marketing activities through the social media and it is expected that it will bring a foundation of establishing smart business strategies by using the network indicators.