| 神川 怜史 | M, 2回目発表 | 計算システムズ生物学 | 金谷 重彦, 佐藤 嘉伸, 佐藤 哲大（客員） |
|
title: An analysis of eye movements during medical image interpretation
abstract: Medical image interpretation is not easy skill for trainees in a limited training period with highly specialization and advance. To clarify the skill, studies have mainly focused on static analysis of gaze time and frequency and the analysis is only possible after whole eye movement data acquisition. In our study, we propose the analysis of fixation tremor (Fixation) and jumping motion (Saccade) using deep learning models and compare 3 different modeling methods. Unlike the conventional method, convolutional neural network methods can be applied easily to the real time analysis during eye movement data acquisition. Tobii Pro X3-120 with a sampling frequency of 120 Hz was used. A radiological technologist with professional experience was participated and there were chest X-ray image datasets used in the experiment. For making the training dataset, the eye movement data was divided into the sections of fixation and saccade by a conventional method, “gazepath”, a package of “R” statistical software. As machine learning methods, we compare two different networks such as transfer learning using MNIST datasets and self-made deep convolutional neural network (DCNN). As a result, the test data accuracy was better for DCNN model than transfer learning using MNIST datasets. language of the presentation: Japanese | |||
| 城戸 孝士郎 | D, 中間発表 | 計算システムズ生物学 | 金谷 重彦, 佐藤 嘉伸, 黄 銘, 小野 直亮, MD.ALTAF-UL-AMIN |
|
title: CNN-Based Classification of Signal Quality and Sleep Position from a Capacitive ECG Measurement for Heart Health Monitoring
abstract: In this study, we focus on the development of a heart health monitoring system based on capacitive electrocardiographs (cECG), which is an unconstrained way of measuring the ECG. cECG is implemented by the electrodes placed on the bed which can measure the ECG signal by capacitive coupling. Therefore, the cECG has a unparalleled advantage over the traditional ECG measurement and can be measured during sleep every day. However, cECG may have unstable waveforms during sleep position changes or even body movements, which is an obstacle to heart monitoring. Therefore, we aim to realize heart health monitoring by automatically recognizing unstable signal fluctuations of cECG using machine learning. language of the presentation: Japanese | |||
| 津曲 優子 | D, 中間発表 | 計算システムズ生物学 | 金谷 重彦, 佐藤 嘉伸, MD.ALTAF-UL-AMIN, 小野 直亮, 黄 銘 |
|
title: Transition of ability of daily activity in hospitalized elderly -Effects of horticultural therapy
abstract: In psychiatric hospitals, prolonged hospitalization and aging are problems. Patients has few events and little role in inpatient life, which may lead to a decline in cognitive function and activity of daily living (ADL). Thereby further extending the length of hospitalization. As one of the countermeasures, horticultural therapy and music therapy is carried out. In this presentation, we will describe the results so far and work plans from now on. language of the presentation: Japanese 発表題目: 入院高齢者の日常生活動作能力の推移-農園活動の効果 発表概要: 精神科病院では入院の長期化，高齢化が問題となっている．入院生活は，変化が少なく，役割も少ないため認知機能の低下や，それによる日常生活活動(ADL)の低下を助長し，更なる在院日数の延長を招きかねない．我々は，その対策の1つとして農園活動や音楽活動を実施している．本発表では，これまでの検証結果と今後の計画について述べる． | |||