Abstract

Detecting pre-cancer in epithelial tissues such as the cervix is a challenging task in low-resources settings. In an effort to achieve low cost cervical cancer screening and diagnostic method for use in low resource settings, mobile colposcopes that use a smartphone as their engine have been developed. Designing image analysis software suited for this task requires proper modeling of light propagation from the abnormalities inside tissues to the camera of the smartphones. Different simulation methods have been developed in the past, by solving light diffusion equations, or running Monte Carlo simulations. Several algorithms exist for the latter, including MCML and the recently developed MCX. For imaging purpose, the observable parameter of interest is the reflectance profile of a tissue under some specific pattern of illumination and optical setup. Extensions of the MCX algorithm to simulate this observable under these conditions were developed. These extensions were validated against MCML and diffusion theory for the simple case of contact measurements, and reflectance profiles under colposcopy imaging geometry were also simulated. To validate this model, the diffuse reflectance profiles of tissue phantoms were measured with a spectrometer under several illumination and optical settings for various homogeneous tissues phantoms. The measured reflectance profiles showed a non-trivial deviation across the spectrum. Measurements of an added absorber experiment on a series of phantoms showed that absorption of dye scales linearly when fit to both MCX and diffusion models. More work is needed to integrate a pupil into the experiment.

© 2015 SPIE