An Optimal Finite Element Error Estimate for an Inverse Problem in Multispectral Bioluminescence Tomography
Rongfang Gong,Xiaoliang Cheng
DOI: https://doi.org/10.1093/imamat/hxt031
2013-01-01
IMA Journal of Applied Mathematics
Abstract:Inspired by the paper (Gong, W., Li, R., Yan, N. N. & Zhao, W. B. (2008) An improved error analysis for finite element approximation of bioluminescence tomography. J. Comput. Math., 26, 1-13), we consider an optimal finite element error estimate for an inverse problem in multispectral bioluminescence tomography. Different from achromatic or monochromatic measurements, hyperspectral or multispectral data can reduce the ill-posedness of the inverse problem and yield improved depth reconstruction. Compared with Gong et al. (2008, An improved error analysis for finite element approximation of bioluminescence tomography. J. Comput. Math., 26, 1-13), which just improved the error order of piecewise constant light source function, error order in this paper is optimal and all error estimates here are valid for a general smooth domain rather than a polyhedral/polygonal one. Moreover, under a boundedness assumption for the admissible source set, the constants in our error estimates do not depend on the regularization parameter, and therefore are bounded.