Nonlinear color segmentation of optical diffusion tomograms reconstructed by the photon average trajectory method

摘要

The photon average trajectory method has been recently investigated as a fast reconstruction technique for time-domain diffuse optical tomography. The main disadvantage of this method is that it reconstructs the tomograms blurred due to averaging over the spatial distributions of photons. To get information about actual boundary and shape of optical inhomogeneities being reconstructed, we propose the segmentation approach based on the generation of nonlinear analytical and statistical functions of correspondence between image intensity and color space. It is shown that for simple models (absorbing macro-inhomogeneities in a homogeneous scattering medium) the proposed approach allows the true structure of inhomogeneities to be reproduced almost completely. If a medium contains randomly inhomogeneous component, our segmentation method may give artifacts which should be removed on the basis of a priori information.