Abstract

Since the linear equation systems that have to be solved during reconstruction in 3D spectral diffuse optical tomography are very large, a direct inversion of the matrix is not feasible. Alternatively iterative solving methods can be used. ART is an iterative method that is character ized by low memory requirements and low computation time for one iteration step, but also low speed of convergence. We present modifications for ART that speed up the convergence espe cially for spectral reconstruction significantly.

© 2008 Optical Society of America