• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.2
• /
• pp.67-72
• /
• 2022

Recently, various intelligent application services using artificial intelligence are being actively developed. In particular, research on artificial intelligence-based real-time prediction services is being actively conducted in the manufacturing industry, and the demand for artificial intelligence services that can detect and predict fire and odors is very high. However, most of the existing detection and prediction systems do not predict the occurrence of fires and odors, but rather provide detection services after occurrence. This is because AI-based prediction service technology is not applied in existing systems. In addition, fire prediction, odor detection and odor level prediction services are services with ultra-low delay characteristics. Therefore, in order to provide ultra-low-latency prediction service, edge computing technology is combined with artificial intelligence models, so that faster inference results can be applied to the field faster than the cloud is being developed. Therefore, in this paper, we propose an LSTM algorithm-based learning model that can be used for fire prediction and odor detection/prediction, which are most required in the manufacturing industry. In addition, the proposed learning model is designed to be implemented in edge devices, and it is proposed to receive real-time sensor data from the IoT terminal and apply this data to the inference model to predict fire and odor conditions in real time. The proposed model evaluated the prediction accuracy of the learning model through three performance indicators, and the evaluation result showed an average performance of over 90%.

• Journal of the Korean Society of Radiology
• /
• v.83 no.2
• /
• pp.331-343
• /
• 2022

Purpose To systematically compare the performance of liver imaging reporting and data system treatment response (LR-TR) with the modified Response Evaluation Criteria in Solid Tumors (mRECIST) for diagnosing viable hepatocellular carcinoma (HCC) treated with locoregional therapy (LRT). Materials and Methods Original studies of intra-individual comparisons between the diagnostic performance of LR-TR and mRECIST using dynamic contrast-enhanced CT or MRI were searched in MEDLINE and EMBASE, up to August 25, 2021. The reference standard for tumor viability was surgical pathology. The meta-analytic pooled sensitivity and specificity of the viable category using each criterion were calculated using a bivariate random-effects model and compared using bivariate meta-regression. Results For five eligible studies (430 patients with 631 treated observations), the pooled per-lesion sensitivities and specificities were 58% (95% confidence interval [CI], 45%-70%) and 93% (95% CI, 88%-96%) for the LR-TR viable category and 56% (95% CI, 42%-69%) and 86% (95% CI, 72%-94%) for the mRECIST viable category, respectively. The LR-TR viable category provided significantly higher pooled specificity (p < 0.01) than the mRECIST but comparable pooled sensitivity (p = 0.53). Conclusion The LR-TR algorithm demonstrated better specificity than mRECIST, without a significant difference in sensitivity for the diagnosis of pathologically viable HCC after LRT.

• Investigative Magnetic Resonance Imaging
• /
• v.12 no.2
• /
• pp.131-141
• /
• 2008

Purpose : To investigate the feasibility and accuracy of Proton Resonance Frequency (PRF) shift based magnetic resonance (MR) temperature mapping utilizing the self-developed center array-sequencing phase unwrapping (PU) method for non-invasive temperature monitoring. Materials and Methods : The computer simulation was done on the PU algorithm for performance evaluation before further application to MR thermometry. The MR experiments were conducted in two approaches namely PU experiment, and temperature mapping experiment based on the PU technique with all the image postprocessing implemented in MATLAB. A 1.5T MR scanner employing a knee coil with $T2^*$ GRE (Gradient Recalled Echo) pulse sequence were used throughout the experiments. Various subjects such as water phantom, orange, and agarose gel phantom were used for the assessment of the self-developed PU algorithm. The MR temperature mapping experiment was initially attempted on the agarose gel phantom only with the application of a custom-made thermoregulating water pump as the heating source. Heat was generated to the phantom via hot water circulation whilst temperature variation was observed with T-type thermocouple. The PU program was implemented on the reconstructed wrapped phase images prior to map the temperature distribution of subjects. As the temperature change is directly proportional to the phase difference map, the absolute temperature could be estimated from the summation of the computed temperature difference with the measured ambient temperature of subjects. Results : The PU technique successfully recovered and removed the phase wrapping artifacts on MR phase images with various subjects by producing a smooth and continuous phase map thus producing a more reliable temperature map. Conclusion : This work presented a rapid, and robust self-developed center array-sequencing PU algorithm feasible for the application of MR temperature mapping according to the PRF phase shift property.

• The Korean Journal of Nuclear Medicine
• /
• v.39 no.6
• /
• pp.445-455
• /
• 2005

Purpose: We developed an animal SPECT system using clinical Philips ARGUS scintillation camera and pinhole collimator with specially manufactured small apertures. In this study, we evaluated the physical characteristics of this system and biological feasibility for animal experiments. Materials and Methods: Rotating station for small animals using a step motor and operating software were developed. Pinhole inserts with small apertures (diameter of 0.5, 1.0, and 2.0 mm) were manufactured and physical parameters including planar spatial resolution and sensitivity and reconstructed resolution were measured for some apertures. In order to measure the size of the usable field of view according to the distance from the focal point, manufactured multiple line sources separated with the same distance were scanned and numbers of lines within the field of view were counted. Using a Tc-99m line source with 0.5 mm diameter and 12 mm length placed in the exact center of field of view, planar spatial resolution according to the distance was measured. Calibration factor to obtain FWHM values in 'mm' unit was calculated from the planar image of two separated line sources. Te-99m point source with i mm diameter was used for the measurement of system sensitivity. In addition, SPECT data of micro phantom with cold and hot line inserts and rat brain after intravenous injection of [I-123]FP-CIT were acquired and reconstructed using filtered back protection reconstruction algorithm for pinhole collimator. Results: Size of usable field of view was proportional to the distance from the focal point and their relationship could be fitted into a linear equation (y=1.4x+0.5, x: distance). System sensitivity and planar spatial resolution at 3 cm measured using 1.0 mm aperture was 71 cps/MBq and 1.24 mm, respectively. In the SPECT image of rat brain with [I-123]FP-CIT acquired using 1.0 mm aperture, the distribution of dopamine transporter in the striatum was well identified in each hemisphere. Conclusion: We verified that this new animal SPECT system with the Phlilps ARGUS scanner and small apertures had sufficient performance for small animal imaging.

• Journal of Internet Computing and Services
• /
• v.14 no.6
• /
• pp.1-8
• /
• 2013

Data mining techniques are used to find important and meaningful information from huge databases, and pattern mining is one of the significant data mining techniques. Pattern mining is a method of discovering useful patterns from the huge databases. Frequent pattern mining which is one of the pattern mining extracts patterns having higher frequencies than a minimum support threshold from databases, and the patterns are called frequent patterns. Traditional frequent pattern mining is based on a single minimum support threshold for the whole database to perform mining frequent patterns. This single support model implicitly supposes that all of the items in the database have the same nature. In real world applications, however, each item in databases can have relative characteristics, and thus an appropriate pattern mining technique which reflects the characteristics is required. In the framework of frequent pattern mining, where the natures of items are not considered, it needs to set the single minimum support threshold to a too low value for mining patterns containing rare items. It leads to too many patterns including meaningless items though. In contrast, we cannot mine any pattern if a too high threshold is used. This dilemma is called the rare item problem. To solve this problem, the initial researches proposed approximate approaches which split data into several groups according to item frequencies or group related rare items. However, these methods cannot find all of the frequent patterns including rare frequent patterns due to being based on approximate techniques. Hence, pattern mining model with multiple minimum supports is proposed in order to solve the rare item problem. In the model, each item has a corresponding minimum support threshold, called MIS (Minimum Item Support), and it is calculated based on item frequencies in databases. The multiple minimum supports model finds all of the rare frequent patterns without generating meaningless patterns and losing significant patterns by applying the MIS. Meanwhile, candidate patterns are extracted during a process of mining frequent patterns, and the only single minimum support is compared with frequencies of the candidate patterns in the single minimum support model. Therefore, the characteristics of items consist of the candidate patterns are not reflected. In addition, the rare item problem occurs in the model. In order to address this issue in the multiple minimum supports model, the minimum MIS value among all of the values of items in a candidate pattern is used as a minimum support threshold with respect to the candidate pattern for considering its characteristics. For efficiently mining frequent patterns including rare frequent patterns by adopting the above concept, tree based algorithms of the multiple minimum supports model sort items in a tree according to MIS descending order in contrast to those of the single minimum support model, where the items are ordered in frequency descending order. In this paper, we study the characteristics of the frequent pattern mining based on multiple minimum supports and conduct performance evaluation with a general frequent pattern mining algorithm in terms of runtime, memory usage, and scalability. Experimental results show that the multiple minimum supports based algorithm outperforms the single minimum support based one and demands more memory usage for MIS information. Moreover, the compared algorithms have a good scalability in the results.

• Journal of Intelligence and Information Systems
• /
• v.26 no.1
• /
• pp.1-21
• /
• 2020

With the rapid development of artificial intelligence technology, various techniques have been developed to extract meaningful information from unstructured text data which constitutes a large portion of big data. Over the past decades, text mining technologies have been utilized in various industries for practical applications. In the field of business intelligence, it has been employed to discover new market and/or technology opportunities and support rational decision making of business participants. The market information such as market size, market growth rate, and market share is essential for setting companies' business strategies. There has been a continuous demand in various fields for specific product level-market information. However, the information has been generally provided at industry level or broad categories based on classification standards, making it difficult to obtain specific and proper information. In this regard, we propose a new methodology that can estimate the market sizes of product groups at more detailed levels than that of previously offered. We applied Word2Vec algorithm, a neural network based semantic word embedding model, to enable automatic market size estimation from individual companies' product information in a bottom-up manner. The overall process is as follows: First, the data related to product information is collected, refined, and restructured into suitable form for applying Word2Vec model. Next, the preprocessed data is embedded into vector space by Word2Vec and then the product groups are derived by extracting similar products names based on cosine similarity calculation. Finally, the sales data on the extracted products is summated to estimate the market size of the product groups. As an experimental data, text data of product names from Statistics Korea's microdata (345,103 cases) were mapped in multidimensional vector space by Word2Vec training. We performed parameters optimization for training and then applied vector dimension of 300 and window size of 15 as optimized parameters for further experiments. We employed index words of Korean Standard Industry Classification (KSIC) as a product name dataset to more efficiently cluster product groups. The product names which are similar to KSIC indexes were extracted based on cosine similarity. The market size of extracted products as one product category was calculated from individual companies' sales data. The market sizes of 11,654 specific product lines were automatically estimated by the proposed model. For the performance verification, the results were compared with actual market size of some items. The Pearson's correlation coefficient was 0.513. Our approach has several advantages differing from the previous studies. First, text mining and machine learning techniques were applied for the first time on market size estimation, overcoming the limitations of traditional sampling based- or multiple assumption required-methods. In addition, the level of market category can be easily and efficiently adjusted according to the purpose of information use by changing cosine similarity threshold. Furthermore, it has a high potential of practical applications since it can resolve unmet needs for detailed market size information in public and private sectors. Specifically, it can be utilized in technology evaluation and technology commercialization support program conducted by governmental institutions, as well as business strategies consulting and market analysis report publishing by private firms. The limitation of our study is that the presented model needs to be improved in terms of accuracy and reliability. The semantic-based word embedding module can be advanced by giving a proper order in the preprocessed dataset or by combining another algorithm such as Jaccard similarity with Word2Vec. Also, the methods of product group clustering can be changed to other types of unsupervised machine learning algorithm. Our group is currently working on subsequent studies and we expect that it can further improve the performance of the conceptually proposed basic model in this study.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_3
• /
• pp.1475-1490
• /
• 2021

Detecting oil spill area using statistical characteristics of SAR images has limitations in that classification algorithm is complicated and is greatly affected by outliers. To overcome these limitations, studies using neural networks to classify oil spills are recently investigated. However, the studies to evaluate whether the performance of model shows a consistent detection performance for various oil spill cases were insufficient. Therefore, in this study, two CNNs (Convolutional Neural Networks) with basic structures(Simple CNN and U-net) were used to discover whether there is a difference in detection performance according to the structure of CNN and distribution characteristics of oil spill. As a result, through the method proposed in this study, the Simple CNN with contracting path only detected oil spill with an F1 score of 86.24% and U-net, which has both contracting and expansive path showed an F1 score of 91.44%. Both models successfully detected oil spills, but detection performance of the U-net was higher than Simple CNN. Additionally, in order to compare the accuracy of models according to various oil spill cases, the cases were classified into four different categories according to the spatial distribution characteristics of the oil spill (presence of land near the oil spill area) and the clarity of border between oil and seawater. The Simple CNN had F1 score values of 85.71%, 87.43%, 86.50%, and 85.86% for each category, showing the maximum difference of 1.71%. In the case of U-net, the values for each category were 89.77%, 92.27%, 92.59%, and 92.66%, with the maximum difference of 2.90%. Such results indicate that neither model showed significant differences in detection performance by the characteristics of oil spill distribution. However, the difference in detection tendency was caused by the difference in the model structure and the oil spill distribution characteristics. In all four oil spill categories, the Simple CNN showed a tendency to overestimate the oil spill area and the U-net showed a tendency to underestimate it. These tendencies were emphasized when the border between oil and seawater was unclear.

• Journal of Korea Water Resources Association
• /
• v.48 no.1
• /
• pp.57-67
• /
• 2015

One of main benefits of a dual polarization radar is improvement of quantitative rainfall estimation. In this paper, performance of two representative rainfall estimation methods for a dual polarization radar, JPOLE and CSU algorithms, have been compared by using data from a MOLIT S-band dual polarization radar. In addition, this paper presents evaluation of specific differential phase ($K_{dp}$) retrieval algorithm proposed by Lim et al. (2013). Current $K_{dp}$ retrieval methods are based on range filtering technique or regression analysis. However, these methods can result in underestimating peak $K_{dp}$ or negative values in convective regions, and fluctuated $K_{dp}$ in low rain rate regions. To resolve these problems, this study applied the $K_{dp}$ distribution method suggested by Lim et al. (2013) and evaluated by adopting new $K_{dp}$ to JPOLE and CSU algorithms. Data were obtained from the Mt. Biseul radar of MOLIT for two rainfall events in 2012. Results of evaluation showed improvement of the peak $K_{dp}$ and did not show fluctuation and negative $K_{dp}$ values. Also, in heavy rain (daily rainfall > 80 mm), accumulated daily rainfall using new $K_{dp}$ was closer to AWS observation data than that using legacy $K_{dp}$, but in light rain(daily rainfall < 80mm), improvement was insignificant, because $K_{dp}$ is used mostly in case of heavy rain rate of quantitative rainfall estimation algorithm.

• The Journal of Korean Society for Radiation Therapy
• /
• v.26 no.1
• /
• pp.107-114
• /
• 2014

Purpose : The purpose of this study is evaluation for the applicability of O-MAR(Metal artifact Reduction for Orthopedic Implants)(ver. 3.6.0, Philips, Netherlands) in head & neck radiation treatment planning CT with metal artifact created by dental implant. Materials and Methods : All of the in this study's CT images were scanned by Brilliance Big Bore CT(Philips, Netherlands) at 120kVp, 2mm sliced and Metal artifact reduced by O-MAR. To compare the original and reconstructed CT images worked on RTPS(Eclipse ver 10.0.42, Varian, USA). In order to test the basic performance of the O-MAR, The phantom was made to create metal artifact by dental implant and other phantoms used for without artifact images. To measure a difference of HU in with artifact images and without artifact images, homogeneous phantom and inhomogeneous phantoms were used with cerrobend rods. Each of images were compared a difference of HU in ROIs. And also, 1 case of patient's original CT image applied O-MAR and density corrected CT were evaluated for dose distributions with SNC Patient(Sun Nuclear Co., USA). Results : In cases of head&neck phantom, the difference of dose distibution is appeared 99.8% gamma passing rate(criteria 2 mm / 2%) between original and CT images applied O-MAR. And 98.5% appeared in patient case, among original CT, O-MAR and density corrected CT. The difference of total dose distribution is less than 2% that appeared both phantom and patient case study. Though the dose deviations are little, there are still matters to discuss that the dose deviations are concentrated so locally. In this study, The quality of all images applied O-MAR was improved. Unexpectedly, Increase of max. HU was founded in air cavity of the O-MAR images compare to cavity of the original images and wrong corrections were appeared, too. Conclusion : The result of study assuming restrained case of O-MAR adapted to near skin and low density area, it appeared image distortion and artifact correction simultaneously. In O-MAR CT, air cavity area even turned tissue HU by wrong correction was founded, too. Consequentially, It seems O-MAR algorithm is not perfect to distinguish air cavity and photon starvation artifact. Nevertheless, the differences of HU and dose distribution are not a huge that is not suitable for clinical use. And there are more advantages in clinic for improved quality of CT images and DRRs, precision of contouring OARs or tumors and correcting artifact area. So original and O-MAR CT must be used together in clinic for more accurate treatment plan.

• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.11
• /
• pp.256-264
• /
• 2020

Industrial Wireless Sensor Networks (IWSNs) is exploited to achieve various objectives such as improving productivity and reducing cost in the diversity of industrial application, and it has requirements such as low-delay and high reliability packet transmission. To accomplish the requirement, the network manager performs graph construction and resource allocation about network topology, and determines the transmission cycle and path of each node in advance. However, this network management scheme cannot treat mobile devices that cause continuous topology changes because graph reconstruction and resource reallocation should be performed as network topology changes. That is, despite the growing need of mobile devices in many industries, existing scheme cannot adequately respond to path failure caused by movement of mobile device and packet loss in the process of path recovery. To solve this problem, a network management scheme is required to prevent packet loss caused by mobile devices. Thus, we analyse the location and movement cycle of mobile devices over time using machine learning for predicting the mobility pattern. In the proposed scheme, the network manager could prevent the problems caused by mobile devices through performing graph construction and resource allocation for the predicted network topology based on the movement pattern. Performance evaluation results show a prediction rate of about 86% compared with actual movement pattern, and a higher packet delivery ratio and a lower resource share compared to existing scheme.