SURE regularization for diffuse optical tomography

Diffuse optical tomography is a morphological imaging technique reconstructs the image based on the diffuse propagation of light in soft biological tissues. The optical parameters recovery provides spatial and structural information, due to diffuse light the problem is nonlinear and ill-posed. To overcome the ill-posed problem in Diffuse Optical tomography SURE (Stein´s Unbiased Risk Estimate) Regularization is proposed to optimize the data. SURE is used to select the parameters and estimate the Mean Square Error (MSE) in inverse problems with Gaussian model. Update the Jacobain matrix for fast variant which can accommodate a variety of regularization techniques. We compare the SURE based regularization with Tikhonov Regularization and Exponential Regularization. We demonstrate the numerical simulations for various Regularization techniques and compare it with SURE regularization parameters which yield near MSE optimal result for non-linear image restoration.