Quantitative diffuse optical tomography requires nonlinear reconstruction algorithms

Summary form only given. Many diffuse optical tomography (DOT) algorithms rely on assumptions that linearize the relationship between the optical contrast of an optical inhomogeneity and measurements of diffuse light. In this paper, we show that this linearisation results in incorrect moments and thus accurate quantitative imaging is not possible. Common examples of linearized image reconstruction algorithms for DOT are based on the Born approximation. Quantitative imaging is possible only if the linearized solution produces scattered waves that agree well with exact solutions. To check this, we calculated the Born approximation for a spherical object embedded in an otherwise uniform, infinite medium, and compared the numerical results with the analytic solution for diffuse photon density waves (DPDW) scattering from spherical objects.