| 木村 友香 | D, 中間発表 | 数理情報学 | 池田 和司 | 佐藤 嘉伸 | 稲垣 直之 |
|
title: Pharmacological functional MRI analysis of dopamine D1 receptor interventions with 11.7 T high field MRI scanner
abstract: Dopamine D1 receptor (D1R) is associated with schizophrenia, Parkinson's disease and attention deficit hyperactivity disorder, and attracting attention as an important therapeutic target. However, the neurophysiological function of D1R has not been completely clarified. Pharmacological functional MRI (phfMRI) evaluates regional brain hemodynamic changes brought by the pharmacological interventions and can help to investigate the neurophysiological function of specific receptors. In this study, we investigated the blood oxygenation level dependent (BOLD) effects of pharmacological interventions using compounds specific for D1R with a preclinical high-field MRI scanner. language of the presentation: Japanese 発表題目: MRIを用いたドパミンD1受容体の薬理学的機能評価 発表概要: ドパミンD1受容体は、統合失調症、パーキンソン病、注意欠陥多動性障害に関連しており、重要な治療標的として注目を集めている。しかしながら、ドパミンD1受容体の神経生理学的機能は完全に解明されていない。本研究では、ドパミンD1受容体の機能評価を目的として、特異的な作動薬、拮抗薬を用いてラット脳内における神経活動変化をfunctional MRIを用いて評価した。 | |||||
| 上田 俊太郎 | M, 1回目発表 | インタラクティブメディア設計学 | 加藤 博一 | 浦岡 行治 | 清川 清 | 神原 誠之 | 藤本 雄一郎 |
|
title: Establishment of Technology to Automatically Build Augmented Reality task support systems
abstract:It is known that the time required for work can be short by using the AR task support system.In addition, it is known that the cognitive cost of workers can be reduced because the information that encourages work support can be superposed on the work target. However, AR task support systems are not widespread in society. The main reason is that the quality of the AR task support system is not stable because it depends greatly on the skill of the system creator. Therefore, we propose a system that automatically generates an AR task support system from conventional task instruction books and 3D CAD data.In this presentation, we will report on the procedure and progress of creating an automatic generation system. language of the presentation:Japanese | |||||||
| 児玉 哲哉 | M, 1回目発表 | インタラクティブメディア設計学 | 加藤 博一 | 向川 康博 | 神原 誠之 | 藤本 雄一郎 | |
|
title:Accurate 3D pose estimation with monocular camera and accurate 3D human model in 100m sprint scene
abstract:In the 100m track and field competitions, form analysis is used to improve performance. One of the easiest ways to analyze form is to watch images and videos, but they are limited to two-dimensional data. Therefore, if you can get the 3D pose, you can analyze the form from any direction.Motion capture is a typical method for estimating 3D pose, inconvenient due to the use of markers and special equipment. Therefore, we use an accurate 3D human model and a monocular RGB camera to estimate the 3D pose from the sprint scene. In our method, we estimate 3D poses by fitting a 3D human body model to the human region in the captured image.In this presentation, we describe the algorithm and the challenges of our method. language of the presentation: Japanese | |||||||
| 大山 航平 | M, 1回目発表 | ユビキタスコンピューティングシステム | 安本 慶一 | 清川 清 | 諏訪 博彦 | 松田 裕貴 | 中村優吾 |
|
title: An investigation into the impact of expressing intentions to improve responses reliability in participatory sensing
abstract: In participatory sensing, the responders sometimes aim to complete the task quickly and effortlessly.This attitude, called "satisficing", make reliability of the collected answer worse.In web-based questionnaires, a study showed that the reliability of responses could be improved by introducing a question that asks respondents to state their position before asking a question, "I will answer seriously."In web-based questionnaires, research has shown that the reliability of responses can be improved by introducing questions that encourage people to state their intention to answer seriously before asking a question.Therefore, this study aims to improve the reliability of responses by introducing a method of expressing the respondent's intention in participatory sensing.However, effective methods of participatory sensing has not yet been found because the tasks and situations used in participatory sensing are different from those in web-based surveys.In this presentation, we propose several new methods for expressing opinions using smartphone sensors and report on our system. language of the presentation: Japanese 発表題目: 参加型センシングの回答信頼性向上に向けた立場表明の影響調査 発表概要: 人がデータを手入力する参加型センシングにおいて、楽に早くタスクを完了しようとする「努力の最小限化」という態度により回答の信頼性が低下する問題がある。Webアンケート調査においては、質問の前に「真面目に回答する」と立場表明を促すような設問を導入することで、回答の信頼性が向上する可能性を示した研究結果がある。そこで本研究では、立場表明の仕組みを参加型センシングに導入することで回答の信頼性向上を目指している。しかしながら、Webアンケート調査と参加型センシングでは要求するタスクや状況が異なるため有効な手法が明らかとなっていない。本発表では、スマートフォンのセンシング機構を利用した新たな立場表明法を複数提案するとともに、構築したシステムについて報告する。 | |||||||
| 生坂 優太 | M, 1回目発表 | 光メディアインタフェース | 向川 康博 | 清川 清 | 舩冨 卓哉 | 田中 賢一郎 | |
|
title: Towards Scene Understanding Based on High Time-Resolved Measurements of Light Propagation
abstract: Transient property of light propagation has many ques to understand the scene. However, the propagation of light was difficult to capture with a conventional camera because the speed of light is too fast. It is ignored in the field of traditional computer vison. In this study, we use a single-photon detector that can measure images with a very high time resolution of ten picoseconds. We confirmed to directly observe light propagation in scenes with reflections, scattering, etc. In the future, we aim to improve the ability of computer vision to understand scenes by tracking rays of light following the physical models. language of the presentation: Japanese | |||||||
| 青山 尚正 | M, 1回目発表 | 光メディアインタフェース | 向川 康博 | 清川 清 | 舩冨 卓哉 | 田中 賢一郎 | |
|
title: Shape Measurement of Specular Objects using Single Photon Avalanche Diode abstract: Measuring the shape of specular objects, such as metals, is useful in fields like a precision work. However, it is difficult to measure the shape of specular objects. The reason is that the direction of reflection is uniquely determined for objects with specular reflection. This property makes it difficult to obtain enough spatial information to make measurements. In this study, we propose a method to measure the shape using double pulse with delay, to supplement the lack spatial information with temporal information. We also show the validation of the method in the simulation. language of the presentation: Japanese | |||||||
| 久家 拓也 | M, 1回目発表 | 計算システムズ生物学 | 金谷 重彦 | 安本 慶一 | 小野 直亮 | MD.ALTAF-UL-AMIN | 黄 銘 |
|
title: Automatic detection of CT lung tumor image location using 3D convolutional neural networks
abstract: Lung cancer is the most frequently diagnosed cancer, and low-dose computed tomography (CT) is now an effective method for early detection of lung cancer, but it requires careful interpretation of each image, which is a heavy burden on radiologists. Although the chest CT generates slice volumes showing various volume representations of the lung structure, the 2-dimensional convolution by the convolution neural network (CNN) does not take the depth direction into consideration, so that the 3-dimensional context information cannot be fully utilized. In this study, we examine the binary classification (benign and malignant) and ternary classification (Benign, primary malignant, and metastatic malignant) of CT images of lung tumors using 3DCNN. In addition to this, we apply two types of CNN methods, i.e., (1)the 3D CNN with chain architecture, which consists of alternating convolution and pooling layers, and (2) the 3D CNN with DAG architecture (Kang G et al., PLoS One 12,:e 188290 (2017) ). For the dataset, previous studies used Lung Image Database Consortium image collection (LIDC-IDRI), but this study will use A Large-Scale CT and PET/CT Dataset for Lung Cancer Diagnosis (Lung-PET-CT-Dx). The data of these related research and use schedule are introduced. language of the presentation: Japanese | |||||||