Scholarlink

In the operation of pattern recognition-based myoelectric prostheses, the ability to provide separated muscle contraction patterns for different motions and reproduce them repeatedly is crucial. Therefore, some biofeedback training methods have been developed to facilitate such operations. In this study, a real-time three-dimensional (3D) feature space feedback system was developed. In this system, high-dimensional features extracted from surface electromyography are compressed into three dimensions by dimensionality reduction, and a 3D graph with different colored clusters is generated. The efficacy of the proposed system was evaluated through a comparative experiment with a conventional two-dimensional feedback system. The results demonstrated that three subjects improved their motion reproduction scores using the proposed method. Furthermore, the effectiveness of the system was found to be higher for users with higher spatial cognitive ability. This suggests that the proposed system is an effective training method for improving the operation of myoelectric prosthetic hand manipulation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Training Method with 3D Feature Space Visualization for Pattern Recognition Controlled Myoelectric Prosthetic Hands

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tatsuki Tsujimoto,
  • Yusuke Yamanoi,
  • Hai Jiang,
  • Yoshiko Yabuki,
  • Yinlai Jiang,
  • Hiroshi Yokoi

摘要

In the operation of pattern recognition-based myoelectric prostheses, the ability to provide separated muscle contraction patterns for different motions and reproduce them repeatedly is crucial. Therefore, some biofeedback training methods have been developed to facilitate such operations. In this study, a real-time three-dimensional (3D) feature space feedback system was developed. In this system, high-dimensional features extracted from surface electromyography are compressed into three dimensions by dimensionality reduction, and a 3D graph with different colored clusters is generated. The efficacy of the proposed system was evaluated through a comparative experiment with a conventional two-dimensional feedback system. The results demonstrated that three subjects improved their motion reproduction scores using the proposed method. Furthermore, the effectiveness of the system was found to be higher for users with higher spatial cognitive ability. This suggests that the proposed system is an effective training method for improving the operation of myoelectric prosthetic hand manipulation.