Optical Diffraction Tomography by Use of Optimization and Phase-Based Fidelity Criterion

Abstract

We introduce a method for reconstructing phase objects in optical diffraction tomography (ODT) based on regularized convex optimization with a new phase-based fidelity criterion. The new criterion is necessary because objects of different refractive-index distributions may produce similar diffracted fields (magnitude and principal-phase) on the detection planes. This surjective, but non-injective relation, attributed to the cyclical nature of the phase, makes optimization algorithms using field-based cost functions prone to local minima that degrade reconstructive accuracy-particularly for objects introducing large optical pathlength difference. A phase-based optimization algorithm that avoids this problem has been developed and tested successfully using simulated data from phantoms and experimental data measured from optical fibers.