Optimal Objective Functional Selection for Image Reconstruction in Diffuse Optical Tomography

Diffused optical tomography (DOT) is a powerful noninvasive functional imaging technique. Inter-parameter crosstalk and near source detector artifacts are major source of inaccuracies in DOT images. In this work we investigate the effect of various objective functional definitions and measurement types on performance of image reconstruction algorithm. Special attention is paid to measurement data types appropriate for handling experimental limitations and inaccuracies. We propose a method of selecting optimal objective functional by visualizing objective functionals in two parameter space using single inclusion DOT problem. Using our method we synthesize a new objective functional for our sample DOT problem. The proposed objective functionals provide minimum inter-parameter crosstalk with negligible near source detector artifacts. Limited memory quasi-Newtonian algorithm is used for image reconstruction. Synthetic data is used to demonstrate effect of various objective functionals on image reconstruction and the superiority of the proposed objective functional