抄録
Recently, EIT has been actively studied for clinical application in Europe. Since the current field supplied the electrodes arranged on the body surface is originally 3-dimensional, we must apply the 3-dimensional algorithm. This solution takes too long calculation time to obtain the correct image. If we can treat EIT problem as 2-dimensional problem, the slice image can be obtained in a short time because the electrode arrangement is limited to the slice. Therefore 2-dimensional EIT is suitable for clinical application. However, the artifact due to difference in dimension is appeared. We propose a new algorithm based on Neural Network to obtain slice image without the problems mentioned above. The algorithm is a technique for obtaining the image by learning the relation between the 3-dimensional transfer impedance on the surface of a slice plane and 2-dimensional conductivity distribution on the plane. In this report, through the simulation experiment assumed the multi-medium structure having wide range of conductivity, the validation of our method was carried out. As a result, it was suggested that the slice image can be estimated from 3-dimensional transfer impedance on the slice.
