Analysis of Frontal Cerebral Blood Flow Changes in Hand Grip Force Adjustment Skill Learning

Human beings perform various motion ranging from simple motion to advanced motion. In addition, sensory information is used as a feedback information of the motion state during operations which require fine adjustment. It is necessary for performing advanced motion to acquire the corresponding skills. To learn such skills efficiently, it is necessary to construct a skill learning system. Grasping skill acquisition status is effective for the skill learning system. Currently, there is no other method than evaluating skill acquisition status by using learners’ task scores. However, this method may not be enough to evaluate skill acquisition status. On the other hand, brain has the function of motion commands and sensory acceptance when moving the body. Therefore, it might be possible to sufficiently evaluate skill acquisition status by using the changes of brain activity. In the paper, we measured the changes of cerebral blood flow on learning of force adjustment skill. The obtained data was analyzed by principal component analysis (PCA) and Recurrent Neural Network (RNN). The result of PCA shows that there are some channels which have the changes of the principal component score before and after learning. In the classification by RNN, when the first principal component scores of the cerebral blood flow data before and after learning were used for input data, high classification accuracy was obtained. In addition, higher classification accuracy was obtained when using only channels which have changes in the principal component scores before and after the learning are used as input data, rather than all channels were inputted. From these results, it is shown that characteristics of the prefrontal cerebral blood flow changes according to the motor skill learning.