琉球大学 · 理工学研究科 · 教授

研究キーワードMedical Electronics・生体医工学・医用電子工学・Brain-Computer Interface・Rehabilitation Robotics・Deep Learning・Medical Image Analysis・Assistive Robotics

ここは公開データのプレビューです。個別の適合理由、連絡角度、保存機能はログイン後に利用できます。

• 理学療法士養成課程における早期臨床実習の経験と学習の分析2026 · Early clinical exposure (ECE) is a practice in which newly enrolled physicians and co-medical students without medical expertise are given opportunities to interact with real patients in the classroom, in the community, and hospital, and its educational effects in improving motivation and emotionality have been reported. However, ECE in hospital facilities, which is highly demanded by students, has not been sufficiently studied because it is difficult for training school educators to observe the content of the practical training and the student's reactions. In this study, we analyzed the context of learning and the experience of practical training from first-year physical therapy students' written reflections developed based on Kolb’s experiential learning theory. The analysis of simple regression revealed that assisting with meals, diaper changing, rehabilitation, and changing positions were useful experiences, and the results of text mining suggested that assisting with meals and baths were useful experiences that enabled learning in the affective domain.
• Development of a CNN-based decision support system for lung disease diagnosis using chest radiographs2025 · Chest radiographs, or chest X-rays (CXRs), are widely used as first-line diagnostic tools for detecting various chest diseases. However, accurately interpreting CXRs remains challenging, as human diagnostic performance is influenced by individual expertise and other factors, often resulting in delays, high costs, and potential misinterpretations. To address these limitations, automated computer-based detection systems offer the potential to enhance diagnostic accuracy, reduce costs, and enable timely disease identification. This study presents CXRNet, a novel, efficient convolutional neural network (CNN)-based framework designed for multi-class classification of common chest diseases, including cardiomegaly, COVID-19, pneumonia, tuberculosis, and normal. The proposed CXRNet is a 16-layer architecture trained on frontal CXR images collected from diverse sources to ensure robust generalization across datasets. The model incorporates advanced strategies to overcome the limitations of previous approaches. Extensive testing under three different data distribution conditions demonstrated the model’s superior performance, achieving an average accuracy of 95.7%, precision of 95.3%, recall of 95.3%, and an F1-score of 95.3% for multi-class classification. Furthermore, for binary classification tasks, CXRNet achieved over 98% average accuracy across all conditions, outperforming existing methods. These results highlight the potential of CXRNet as a reliable decision support system for efficient and accurate chest disease diagnosis, paving the way for real-time clinical applications.
• Development of proning pose classification system for selfcare of COVID-19 patients2025 · The COVID-19 situation is one of the most critical pandemics in human history. In COVID-19, severe breathing problems are caused in most cases due to lung damage that can be minimized by proning, a special type of exercise to increase the oxygen saturation level. The main purpose of this research is then to develop a proning pose classification system for self-care of COVID-19 patients. Since improper proning postures may lead to in juries and fatigue then the proning under the supervision of an expert is highly recommended which is very challenging to accommodate during isolation. To overcome this situation, K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), Convolutional Neural Network (CNN), and Deep Neural Network (DNN) based systems have been developed to recognize the human pose. Additionally, cosine similarity is also estimated to assess the preciseness of the desired pose. MediaPipe library is used to generate 33 key points of the body parts which were used to recognize the human body structure and orientation from real-time poses captured in a web camera or mobile camera. A patient can perform his proning exercise using our developed system with the help of an Android mobile or computer. The SVM-based system with a linear kernel has achieved the best performance, with accuracy (train: 100 %; test: 99.5 %; validation: 99.2 %), precision (99.5 %), recall (99.6 %), F1-score (99.5 %), and AUC (99.7 %). It has also demonstrated the lowest latency of 112 ms, compared to 124 ms for CNN and 122 ms for DNN, respectively. Consequently, this model is recommended as the optimal solution for COVID-19 patients, enabling accurate and safe proning exercises to enhance recovery.
• 重度障がい者を対象とした食事支援ロボットアーム －ペットボトルを用いた水分摂取課題の検証－2025 · To assist people with severe disabilities to eat and drink at their own pace, we present a self-feeding robotic arm system with the functions of object detections using YOLOv5 model and user’s mouth detection using MediaPipe. The redundant robotic arm with 7-DOF (degree-of-freedom) was made to perform natural eating motion and avoid obstacles by considering the home care and nursing care environments in Japan. An open-source software, ROS (Robot Operating System), and a motion planning framework, MoveIt! simulator including the motion planning solver using inverse kinematics, were used to simulate the 7-DOF robotic arm. Moreover, a single-finger operated interface was applied as a controller for the robotic arm system. We demonstrated that using the proposed robotic arm system, the tasks to grasp and move the plastic bottle to the user’s mouth were conducted. From the simulation and experimental results, we found that the detections of the target object and the user’s mouth were conducted effectively. In addition, it was shown that the success rates of the tasks were more than 80% or equal when having no object in the calculated trajectory of the robotic arm. Future works include conducting some experiments of the tasks with people with disabilities.

• XR技術を応用したセルフトレーニングシステム構築と教育効果の検証2024 · 本研究ではXR（クロスリアリティ）技術を用いたシミュレーショントレーニング環境を構築し、遠隔でのセルフトレーニング、チームトレーニングが可能なオンライン教育環境におけるその有効性を検証する。より効果的なオンライントレーニング環境の開発は、withコロナ時代にまた、沖縄県のような島嶼県における離島へき地教育には必要不可欠であり、より実効性のある効果的教育手法の開発という喫緊の課題に寄与することが期待できる。
• 半側空間無視症状検査システムの開発とそのリハビリテーションへの応用2024 · 本研究の目的は、脳卒中の後遺症として知られる半側空間無視症状を有する患者の支援と効果的なリハビリテーションを検討することである。この目的を達成するために、次のことを実施する。 1. 患者の症状を定量的に評価して情報提供を行う検査システムを検討する。 2. 患者の上部体幹系(上肢、頭部、体幹を含む)における姿勢のリハビリテーションを行うための手法を検討する。
• ニューロリハビリテーションを目的とした上肢運動支援システムの開発2019 · 基盤研究(C)
• 食事支援モバイルアーム用インタフェースの開発2013 · 基盤研究(C)

• 公式プロフィール