感度適応型正則化を用いた深さ選択性拡散光イメージング

概要

論文の詳細を見る
In this study, we propose an inverse algorithm that suppresses the undesirable effects of skin circulation in near-infrared diffuse optical topography, which not only provides 2D images of deep regions but also displays undesirable changes in skin circulation. For this algorithm, a voxel is placed under each optode in a shallow plane, and a 2D voxel array is prepared as a target-imaging plane in a deep region. A sensitivity matrix having elements that correspond to these voxels in the shallow plane and target plane is calculated on the basis of optical diffusion approximation. The estimated relative absorption change is obtained from the observation data by Moore-Penrose inverse of the sensitivity matrix. The performances of the following two types of regularization are compared: simple regularization and sensitivity adaptive regularization. Both methods can suppress undesirable noise caused by absorption changes in shallow regions; in particular, the sensitivity adaptive method provides satisfactory reconstructed images of the target region. The performances of the voxels in the shallow plane are also investigated by interpolating the estimated absorption changes corresponding to the voxels to form a 2D image. The result shows that the voxels can provide distinct information about the skin circulation. A phantom experiment also shows the advantages of the proposed method.
一般社団法人日本生体医工学会の論文