抄録
Recently, technology of brain activity measurement has developed, and many researchers have tried to elucidate the human brain function. In those technology, Near Infrared Spectroscopy (NIRS) enables us to measure brain activity safely and easily. That is the reason that NIRS is expected to apply variable study field. Since the spatial resolution of NIRS is not sufficient for further measurement of brain function, which is the disadvantage of NIRS and prevent development of applied research. In some previous researches, the spatial resolution improvement of NIRS have been performed by the minimum norm method and so on. The approaches can three-dimensionally estimate activation site in brain. However, the activation site estimated by the minimum norm method widely spreads, and it is difficult to perform the activity estimation in deep area. Therefore, the purpose of our study is to improve the measurement of activation site in deep position in brain by the modified minimum norm method. In this study, an experimental model based on the head structure is used, and the sensitivity distribution in the model is calculated by Monte Carlo simulation. We focus on the sensitivity distribution, and voxels in the model are classified by some levels depending on intensity of the sensitivity distribution. By comparing estimation value in each level, inactive voxels were determined and estimated active position was able to be narrowed down. As a result, estimation of brain activity in local position is improved compared with the conventional minimum norm method especially in estimation of small active position.
