The prevalence of Dry Eye Disease (DED) in the USA is approximately 40 million in aging adults with about $3.8
billion economic burden. However, a comprehensive understanding of tear film dynamics, which is the prerequisite to
advance the management of DED, is yet to be realized. To extend our understanding of tear film dynamics, we
investigate the simultaneous estimation of the lipid and aqueous layers thicknesses with the combination of optical
coherence tomography (OCT) and statistical decision theory.
In specific, we develop a mathematical model for Fourier-domain OCT where we take into account the different
statistical processes associated with the imaging chain. We formulate the first-order and second-order statistical
quantities of the output of the OCT system, which can generate some simulated OCT spectra. A tear film model, which
includes a lipid and aqueous layer on top of a rough corneal surface, is the object being imaged. Then we further
implement a Maximum-likelihood (ML) estimator to interpret the simulated OCT data to estimate the thicknesses of
both layers of the tear film. Results show that an axial resolution of 1 μm allows estimates down to nanometers scale.
We use the root mean square error of the estimates as a metric to evaluate the system parameters, such as the tradeoff
between the imaging speed and the precision of estimation. This framework further provides the theoretical basics to
optimize the imaging setup for a specific thickness estimation task.