• Korean Journal of Radiology
• /
• 제21권12호
• /
• pp.1345-1354
• /
• 2020

Objective: To evaluate radiomics analysis in studies on mild cognitive impairment (MCI) and Alzheimer's disease (AD) using a radiomics quality score (RQS) system to establish a roadmap for further improvement in clinical use. Materials and Methods: PubMed MEDLINE and EMBASE were searched using the terms 'cognitive impairment' or 'Alzheimer' or 'dementia' and 'radiomic' or 'texture' or 'radiogenomic' for articles published until March 2020. From 258 articles, 26 relevant original research articles were selected. Two neuroradiologists assessed the quality of the methodology according to the RQS. Adherence rates for the following six key domains were evaluated: image protocol and reproducibility, feature reduction and validation, biologic/clinical utility, performance index, high level of evidence, and open science. Results: The hippocampus was the most frequently analyzed (46.2%) anatomical structure. Of the 26 studies, 16 (61.5%) used an open source database (14 from Alzheimer's Disease Neuroimaging Initiative and 2 from Open Access Series of Imaging Studies). The mean RQS was 3.6 out of 36 (9.9%), and the basic adherence rate was 27.6%. Only one study (3.8%) performed external validation. The adherence rate was relatively high for reporting the imaging protocol (96.2%), multiple segmentation (76.9%), discrimination statistics (69.2%), and open science and data (65.4%) but low for conducting test-retest analysis (7.7%) and biologic correlation (3.8%). None of the studies stated potential clinical utility, conducted a phantom study, performed cut-off analysis or calibration statistics, was a prospective study, or conducted cost-effectiveness analysis, resulting in a low level of evidence. Conclusion: The quality of radiomics reporting in MCI and AD studies is suboptimal. Validation is necessary using external dataset, and improvements need to be made to feature reproducibility, feature selection, clinical utility, model performance index, and pursuits of a higher level of evidence.

• 한국초전도ㆍ저온공학회논문지
• /
• 제24권3호
• /
• pp.24-29
• /
• 2022

One of the recent proposals with nuclear magnetic resonance (NMR) is a multi-bore NMR which consists of array of magnets which could present possibilities to quickly cope with pandemic virus by multiple inspection of virus samples. Low temperature superconductor (LTS) can be a candidate for mass production of the magnet due to its low price in fabrication as well as operation by applying the helium zero boil-off technology. However, training feature of LTS magnet still hinders the low cost operation due to multiple boil-offs during premature quenches. Thus in this paper, LTS magnet with low mechanical stress is designed targeting the "training-free" LTS magnet for mass production of magnet array for multi-bore NMR. A thorough process of an LTS magnet design is conducted, including the analyses as the followings: electromagnetics, mechanical stress, cryogenics, stability, and protection. The magnet specification was set to 4.7 T in a winding bore of 85 mm, corresponding to the MR frequency of 200 MHz. The stress level is tolerable with respect to the wire yield strength and epoxy crack where mechanical disturbance is less than the minimum quench energy.

• Journal of Electrical Engineering and Technology
• /
• 제13권5호
• /
• pp.1986-1995
• /
• 2018

In this paper, a novel neutral-point voltage balancing scheme for NPC three-level inverters using carrier-based sinusoidal pulse width modulation (SPWM) method is developed. The new modulation approach, based on the obtained expressions of zero sequence voltage in all six sectors, can significantly suppress the low-frequency voltage oscillation in the neutral point at high modulation index and achieve a fast voltage-balancing dynamic performance. The implementation of the proposed method is very simple. Another attractive feature is that the scheme can stably control any voltage difference between the two dc-link capacitors within a certain range without using any extra hardware. Furthermore, the presented scheme is also applicable to the single-phase NPC three-level inverter. It can maintain the neutral-point voltage balance at full modulation index and improve the voltage-balancing dynamic performance of the single-phase NPC three-level inverter. The performance of the proposed strategy and its benefits over other previous techniques are verified experimentally.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1997년도 한국자동제어학술회의논문집; 한국전력공사 서울연수원; 17-18 Oct. 1997
• /
• pp.708-711
• /
• 1997

This paper proposes the hybrid algorithm for the optimization of the structure and parameters of the fuzzy neural networks by genetic algorithms (GA) to improve the behaviour and the design of fuzzy neural networks. Fuzzy neural networks have a distinguishing feature in that they can possess the advantage of both neural networks and fuzzy systems. In this way, we can bring the low-level learning and computational power of neural networks into fuzzy systems and also high-level, human like IF-THEN rule thinking and reasoning of fuzzy systems into neural networks. As a result, there are many research works concerning the optimization of the structure and parameters of fuzzy neural networks. In this paper, we propose the hybrid algorithm that can optimize both the structure and parameters of fuzzy neural networks. Numerical example is provided to show the advantages of the proposed method.

• 한국농업기계학회:학술대회논문집
• /
• 한국농업기계학회 1998년도 하계 학술대회 논문집
• /
• pp.270-275
• /
• 1998

It was assumed that the voice of cattle is generated with the virtual white noise through the digital filter called the linear prediction filter, and filter parameters (prediction coefficients) were estimated by the maximum entropy method (MEM) , using the sound signal of the animal . The feature planes were defined by the pairs of two parameters selected appropriately from these parameters. The cattle voices were divided into three levels, that is the high, medium and low levels according to their total power equivalent to the variances of the sound signal . It was found that the straight lines could be used for recognizing tow cow and one calf for high level voices. For high and medium level voices, however, it was difficult or impossible to recognize individual cattle on the parameters planes.

• Bulletin of the Korean Chemical Society
• /
• 제32권7호
• /
• pp.2199-2202
• /
• 2011

The electronic properties of altretamine(2,4,6-tris [dimethylamino]-1,3,5-triazine) adsorbed on epitaxial graphene (EG) were investigated by core-level photoemission spectroscopy (CLPES) in conjunction with low energy electron diffraction (LEED). We found that altretamine molecule adsorbed onto interface layer (S1) of graphene as we confirm decrement of S1 peak using CLPES and haziness of LEED pattern. Moreover, the measured work function changes verified that increased adsorption of the altretamine on graphene layer showed n-type doping characteristics due to charge transfer from altretamine to graphene through the nitrogens. Two distinct nitrogen bonding feature associated with the N 1s peak was clearly observed in the core-level spectra indicating two different chemical environments.

• 반도체디스플레이기술학회지
• /
• 제5권3호
• /
• pp.11-16
• /
• 2006

A low cost PDP sustain power supply is proposed based on flyback topology. By using Boundary Conduction Method(BCM) to control input current regulation, DCM condition can be met under all load conditions. Another feature of the proposed method is that a excessive voltage stress due to the link voltage increase can be suppressed by removing link capacitor and suggest new 'Level-shifting switch driver'. this new gate driver is improved 66% of efficiency than switching loss of a existed push-pull amplifier. The proposed converter is tested with a 400W(200V-2A output) prototype circuit.

• ETRI Journal
• /
• 제29권5호
• /
• pp.700-702
• /
• 2007

We present a relevance feedback approach based on multi-class support vector machine (SVM) learning and cluster-merging which can significantly improve the retrieval performance in region-based image retrieval. Semantically relevant images may exhibit various visual characteristics and may be scattered in several classes in the feature space due to the semantic gap between low-level features and high-level semantics in the user's mind. To find the semantic classes through relevance feedback, the proposed method reduces the burden of completely re-clustering the classes at iterations and classifies multiple classes. Experimental results show that the proposed method is more effective and efficient than the two-class SVM and multi-class relevance feedback methods.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2012년도 춘계학술발표대회
• /
• pp.1176-1179
• /
• 2012

To ensure the correctness of high performance satisfiability (SAT) solvers, several proof formats have been proposed. SAT solvers can report a formula being unsatisfiable with a proof, which can be independently verified by a trusted proof checker. Among the proof formats accepted at the SAT competition, the Reverse Unit Propagation (RUP) format is considered the most popular. However, the official proof checker was not efficient and failed to check many of the proofs at the competition. This inefficiency is one of the drawbacks of SAT proof checking. In this paper, I introduce a work-in-progress project, vercheck to implement an efficient RUP checker using modern SAT solving techniques. Even though my implementation is larger and more complex, the level of trust is preserved by statically verifying the correctness of the code. The vercheck program is written in GURU, a dependently typed functional programming language with a low-level resource management feature.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권11호
• /
• pp.3493-3506
• /
• 2022

In the field of scene segmentation, the precise segmentation of object boundaries in sports movement scene images is a great challenge. The geometric information and spatial information of the image are very important, but in many models, they are usually easy to be lost, which has a big influence on the performance of the model. To alleviate this problem, a parallel dense dilated convolution merging Network (termed PDDCM-Net) was proposed. The proposed PDDCMNet consists of a feature extractor, parallel dilated convolutions, and dense dilated convolutions merged with different dilation rates. We utilize different combinations of dilated convolutions that expand the receptive field of the model with fewer parameters than other advanced methods. Importantly, PDDCM-Net fuses both low-level and high-level information, in effect alleviating the problem of accurately segmenting the edge of the object and positioning the object position accurately. Experimental results validate that the proposed PDDCM-Net achieves a great improvement compared to several representative models on the COCO-Stuff data set.