Adaptive selection of minimally correlated data for optimization of source-detector configuration in diffuse optical tomography

Abstract

The optimization of experimental design prior to deployment, not only for cost effective solution but also for computationally efficient image reconstruction has taken up for this study. We implemented the iterative method also known as effective independence (EFI) method for optimization of source/detector pair configuration. The notion behind for adaptive selection of minimally correlated measurements was to evaluate the information content passed by each measurement for estimation of unknown parameter. The EFI method actually ranks measurements according to their contribution to the linear independence of unknown parameter basis. Typically, to improve the solvability of ill conditioned system, regularization parameter is added, which may affect the source/detector selection configuration. We show that the source/detector pairs selected by EFI method were least prone to vary with sub optimal regularization value. Moreover, through series of simulation studies we also confirmed that sparse source/detector pair measurements selected by EFI method offered similar results in comparison with the dense measurement configuration for unknown parameters qualitatively as well as quantitatively. Additionally, EFI method also allow us to incorporate the prior knowledge, extracted in multimodality imaging cases, to design source/detector configuration sensitive to specific region of interest. The source/detector ranking method was further analyzed to derive the automatic cut off number for iterative scheme.