• 한국어정보학회:학술대회논문집
• /
• 2016.10a
• /
• pp.176-180
• /
• 2016

대화형 시스템이 사람의 경청 기술을 모방할 수 있다면 대화 상대방과 더 효과적으로 상호작용 할 수 있을 것이다. 본 논문에서는 시스템이 경청 기술을 모방할 수 있도록 사용자의 발화를 기반으로 질문을 생성하는 것에 대해 연구하였다. 그리고 이러한 연구를 위해 필요한 데이터를 Image captioning과 Visual QA 데이터를 기반으로 생성하고 활용하는 방안에 대해 제안한다. 또한 이러한 데이터를 Attention 메커니즘을 적용한 Sequence to sequence 모델에 적용하여 질문을 생성하고, 생성된 질문의 질문 유형을 분석하였다. 마지막으로 사람이 작성한 질문과 모델의 질문 생성 결과 비교를 BLEU 점수를 이용하여 수행하였다.

• Annual Conference on Human and Language Technology
• /
• 2016.10a
• /
• pp.176-180
• /
• 2016

대화형 시스템이 사람의 경청 기술을 모방할 수 있다면 대화 상대방과 더 효과적으로 상호작용 할 수 있을 것이다. 본 논문에서는 시스템이 경청 기술을 모방할 수 있도록 사용자의 발화를 기반으로 질문을 생성하는 것에 대해 연구하였다. 그리고 이러한 연구를 위해 필요한 데이터를 Image captioning과 Visual QA 데이터를 기반으로 생성하고 활용하는 방안에 대해 제안한다. 또한 이러한 데이터를 Attention 메커니즘을 적용한 Sequence to sequence 모델에 적용하여 질문을 생성하고, 생성된 질문의 질문 유형을 분석하였다. 마지막으로 사람이 작성한 질문과 모델의 질문 생성 결과 비교를 BLEU 점수를 이용하여 수행하였다.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.305-308
• /
• 2002

More complex radiotherapy techniques using multi leaf collimation(MLC) such as intensity-modulated radiation therapy(IMRT) has been increasing the significance of verification of leaf position and motion. Due to the reliability and robustness, quality assurance(QA) of MLC is usually performed with portal films. However, the advantage of ease of use and capability of providing digital data of electronic portal imaging devices(EPIDs) have attracted many attentions as alternatives of films for routine quality assurance in spite of the concerns about their clinical feasibility, efficacy, and the cost to benefit ratio. In our work, the method of routine QA of MLC using electronic portal imaging(EPI) was developed. The verification of availability of EPI images for routine QA was performed by comparison with those of the portal films which were simultaneously obtained when radiation was delivered and known prescription input to MLC controller. Specially designed test patterns of dynamic MLC were applied to image acquisition. Quantitative off-line analysis using edge detection algorithm enhanced the verification procedure in addition to on-line qualitative visual assessment. In conclusion, the EPI is available enough for routine QA with the accuracy of portal films.

• Progress in Medical Physics
• /
• v.14 no.3
• /
• pp.151-160
• /
• 2003

The application of more complex radiotherapy techniques using multileaf collimation (MLC), such as 3D conformal radiation therapy and intensity-modulated radiation therapy (IMRT), has increased the significance of verifying leaf position and motion. Due to thier reliability and empirical robustness, quality assurance (QA) of MLC. However easy use and the ability to provide digital data of electronic portal imaging devices (EPIDs) have attracted attention to portal films as an alternatives to films for routine qualify assurance, despite concerns about their clinical feasibility, efficacy, and the cost to benefit ratio. In this study, we developed method for daily QA of MLC using electronic portal images (EPIs). EPID availability for routine QA was verified by comparing of the portal films, which were simultaneously obtained when radiation was delivered and known prescription input to MLC controller. Specially designed two-test patterns of dynamic MLC were applied for image acquisition. Quantitative off-line analysis using an edge detection algorithm enhanced the verification procedure as well as on-line qualitative visual assessment. In conclusion, the availability of EPI was enough for daily QA of MLC leaf position with the accuracy of portal films.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.12
• /
• pp.485-496
• /
• 2018

So far, the deep learning, a field of artificial intelligence, has achieved remarkable results in solving problems from unstructured data. However, it is difficult to comprehensively judge situations like humans, and did not reach the level of intelligence that deduced their relations and predicted the next situation. Recently, deep neural networks show that artificial intelligence can possess powerful relational reasoning that is core intellectual ability of human being. In this paper, to analyze and observe the performance of Relation Networks (RN) among the neural networks for relational reasoning, two types of RN-based deep neural network models were constructed and compared with the baseline model. One is a visual question answering RN model using Sort-of-CLEVR and the other is a text-based question answering RN model using bAbI task. In order to maximize the performance of the RN-based model, various performance improvement experiments such as hyper parameters tuning have been proposed and performed. The effectiveness of the proposed performance improvement methods has been verified by applying to the visual QA RN model and the text-based QA RN model, and the new domain model using the dialogue-based LL dataset. As a result of the various experiments, it is found that the initial learning rate is a key factor in determining the performance of the model in both types of RN models. We have observed that the optimal initial learning rate setting found by the proposed random search method can improve the performance of the model up to 99.8%.

• Proceedings of the KOSOMBE Conference
• /
• v.1998 no.11
• /
• pp.117-118
• /
• 1998

The aim of this work is to develop 3-D stereotactic localization system in order to determine the precise shape, size and location of the lesion in the brain in the field of Stereotactic Radiosurgery(SRS) and neurosurgery using CT/MRI/angiography and multi-purpose QA phantom. The algorithms to obtain a 3-D stereotactic coordinates of the target have been developed, and targets on each CT image were superimposed each other on MR/angiography images without distortion corretion. This system was implented in Visual C++ as a PC-based application program.

• Korean Journal of Digital Imaging in Medicine
• /
• v.10 no.1
• /
• pp.21-28
• /
• 2008

Diagnostic display monitor QA according to AAPM TG18 is usually performed by PACS administrator, product manager and reading doctor, and for acceptance testing and periodic quality control evaluation, a combination of visual and quantitative tests can be used, as outlined in sections 5 and 6 of 'assessment of display performance for medical imaging systems'. Although many display tests can be performed visually, a more objective and quantitative evaluation of display performance requires special test tools. The required instruments vary in their complexity and cost, depending on the context of the evaluation(research, acceptance testing, or quality control) and how thorough the evaluation needs to be. Objective and reliable assessment of many display characteristics can be performed with relatively inexpensive equipment, So, we made 'AAPM TG18 guiding instrument' to ues variable purpose of the evaluation of 'geometrical distortions(quantitative"', 'veiling glare(visual)' and 'sensor calibration'. The spatial measurements for the quantitative evaluation of geometric distortions, and the measurement of the veling-glare ring response function which provides information regarding the spatial extent of the luminance spread, can be performed using the TG18 guiding instrument can be used to sensor calibration to standardize the basic rate of 0% luminance when periodic calibration.

• Investigative Magnetic Resonance Imaging
• /
• v.3 no.2
• /
• pp.173-178
• /
• 1999

Purpose : To evaluate the usefulness of functional MRI (fMRI) of visual cortex in patients with ischemic infarction in the occipital lobe. Materials and Methods : Four patients with the symptoms and signs of visual cortical ischemia were included. Functional MRI was performed by 2D-FLASH technique with the parameter of 90/56msec TR/TE, $40^{\circ}$ flip angle, $240{\times}240{\;}FOV,{\;}64{\times}128$ matrix number, 8.32 seconds acquisition time, 8mm slice thickness. An axial slice including both visual cortices was selected and alternative activation and resting of the visual cortex was performed using red color photostimulator. all patients undertook visual field test, and vascular abnormality was examined by MRA (n=4) and DSA (n=2). fMRI results were compared with the results of a visual field test, conventional MRI and cerebral angiography. Results : On fMRI, decreased activity of the visual cortex was found in the occipital lobe corresponding to stenosis of the posterior cerebral artery or its branch noted on angiogram. However, 2 of 4 patients showed no abnormal findings on conventional MRI. Visual field defect was noted in 3 patients, one and of whom showed no abnormality on conventional MRI and diffusion-weighted image, but revealed decreased activity in the corresponding visual cortex on fMRI. Conclusion : fMRI may be a sensitive method for detection of the status of decreased blood flow or vascular reserve which other methods can not.

• Progress in Medical Physics
• /
• v.19 no.4
• /
• pp.231-240
• /
• 2008

We developed a user-friendly program to independently verify monitor units (MUs) calculated by radiation treatment planning systems (RTPS), as well as to manage beam database in clinic. The off-axis factor, beam hardening effect, inhomogeneity correction, and the different depth correction were incorporated into the program algorithm to improve the accuracy in calculated MUs. A beam database in the program was supposed to use measured data from routine quality assurance (QA) processes for timely update. To enhance user's convenience, a graphic user interface (GUI) was developed by using Visual Basic for Application. In order to evaluate the accuracy of the program for various treatment conditions, the MU comparisons were made for 213 cases of phantom and for 108 cases of 17 patients treated by 3D conformal radiation therapy. The MUs calculated by the program and calculated by the RTPS showed a fair agreement within ${\pm}3%$ for the phantom and ${\pm}5%$ for the patient, except for the cases of extreme inhomogeneity. By using Visual Basic for Application and Microsoft Excel worksheet interface, the program can automatically generate beam data book for clinical reference and the comparison template for the beam data management. The program developed in this study can be used to verify the accuracy of RTPS for various treatment conditions and thus can be used as a tool of routine RTPS QA, as well as independent MU checks. In addition, its beam database management interface can update beam data periodically and thus can be used to monitor multiple beam databases efficiently.

• Journal of Radiation Protection and Research
• /
• v.44 no.1
• /
• pp.43-52
• /
• 2019

Background: The mid-term performance of clinical linear accelerator (LINAC) during volumetric modulated arc therapy (VMAT) treatment period is not performed in clinical practice and usually replaced with one-time plan quality assurance (QA). In this research we aim to monitor daily reproducibility of VMAT delivery from tracking individual leaf movement error and dosimetric error to evaluate the mid-term quality of the machine used. Materials and Methods: First, multileaf collimator (MLC) information was imported into MATLAB program to determine which of the MLC leaves in the leaf bank had the maximum RMS position error (maxRMS). We estimated where the maximum positional errors (maxPE) of the chosen leaf occur along its path length and tracked its daily variations over the entire treatment period. Secondly, picture information of dosimetric error from portal dosimetry was imported into MATLAB where representative high gamma index region (HGR) was determined as HGR with length of > 1 cm and their centers were daily tracked. Results and Discussion: The maxPEs in the brain and tongue cases were distributed broader than in other cases, but all data were found located within ${\pm}0.5mm$. From first day to last day all of five cases show the similar visual pattern of HGRs and Centers of the longest HGRs remained within ${\pm}1mm$ of that in first day. These findings prove excellent mid-term performance of the LINAC used in VMAT treatments over a full course of treatment. Conclusion: Tracking the daily location changes of leaf movement and dosimetric error can be a good indicator of predicting the daily quality like stability and reproducibility of beam delivering in VMAT treatment.