Reconstructing features of thick objects from phase images

Abstract

Phase microscopy modalities are extensively used to image unstained transparent biological samples because of their ability to obtain high contrast images without exogenous agents. Quantitative phase techniques in particular provide valuable information that can be interpreted easily when the imaged object is optically thin, that is, when the thickness of the object is much less than the depth of field of the imaging system. However, many biological objects of interest have thicknesses comparable to or larger than the depth of field. This work focuses on the initial development of inversion techniques for phase images, in order to reconstruct features of thick transparent samples. We use a shape based iterative approach that assumes that the index of refraction inside the object can be approximated as piecewise constant. The case of thick homogeneous and inhomogeneous objects is examined. We assume that the boundary location of all inhomogeneities is known or can be obtained by preprocessing the image. Our goal is to estimate their unknown indices of refraction. We analyze the performance of the reconstruction when the thickness of the object is increased from 1 to 20 illumination wavelengths (0.633 micrometers). We simulate experiments using a objective lens with a numerical aperture of 0.5. Simulation results for objects with optical properties similar to real transparent biological samples are presented. The reconstructed indices of refraction have an error less than 5% compared to the true value.