ヒト腕等尺性運動タスクにおけるfNIRS信号からの腕の力方向推定

摘要

In this study, the directions of the force exerted by the arm during isometric muscle contraction were estimated by using the 24-channel functional near infrared spectroscopy (fNIRS) signal on cortical hemodynamics. In an experiment, subjects were required to perform isometric arm movements in four directions (Front: F, Right: R, Back: B and Left: L from the subjects) with the right arm while keeping a certain joint angle during the task period. Spatial and temporal feature selection were performed on the fNIRS signal to ascertain the representation of the force directions. The estimation results suggest that the FB-RL classification accuracy was approximately 70-90%. Further, NIRS spatial parametric mapping (NIRS-SPM) analysis results suggest relation to the spatial features of the fNIRS signal which contributed for the estimation. These results indicate the possibility that statistically-significant areas give effective features for estimation the direction on the fNIRS signal.%本報告では，ヒト腕による等尺性運動時のfNIRS信号から運動の方向を推定できるかを検討をし，fNIRS信号からの脳情報推定の可能性を議論する.実験では，被験者が右腕による等尺性筋伸縮運動を，4方向（前，右，後，左）について行い，左半球一次運動野周辺のfNIRS信号を測定した.得られたfNIRS信号に対する，空間的，時間的な特徴選択を行い，推定に寄与する特徴を抽出した.結果として，特徴を適切に選択し，SVMを用いることで，約70-90％の推定精度が得られ，fNIRS信号に，運動方向に関連する情報が表出している可能性を示した.また，NIRS-SPM（NIRS-SpatialParametric Mapping）解析を行うことで統計的に有意な特徴を示し，推定に寄与した特徴量とNIRS-SPM解析の結果に関連があることを示した.