A Monte Carlo simulator of PS-OCT local birefringence imaging (Conference Presentation)

Abstract

Polarization-sensitive optical coherence tomography (PS-OCT) allows imaging of tissue birefringence. In practice, however, PS-OCT images are often confounded by high noise and confusing artifacts. A full understanding of the intrinsic and instrumentation-derived signal and noise properties of PS-OCT has not been developed. In this presentation, we describe a Monte Carlo (MC) simulator of PS-OCT local birefringence imaging that recapitulates the noise and signal properties observed in empirical images and, as such, can be used to understand and improve PS-OCT methods. The MC simulator builds upon a previously described MC methodology that supports arbitrary three-dimensional geometries. To this, we have added support for MC simulation of transverse speckle correlation. This is important because many of the noise sources in PS-OCT are driven by interactions with the speckle field. We have developed a method to support polarization-dependent measurements of birefringent tissues. Both additive (due to finite SNR) and polarization-mode dispersion noise can be incorporated. To demonstrate the utility of the simulator, we use it to reveal a previously unappreciated noise that results from the design of conventional local birefringence extraction algorithms, and we describe an improved method that lowers noise in both simulated and empirical datasets. We anticipate that this simulator will enable new explorations into the fidelity of PS-OCT measurements and accelerate the optimization of PS-OCT methods and algorithms.