Near-infrared frequency domain system and fast inverse Monte Carlo algorithm for endoscopic measurement of tubular tissue

Abstract

A near infrared diffuse reflectance system for endoscopic measurement and an inverse algorithm for extracting optical properties of tubular tissues were developed in this paper. The measurement system worked in the frequency domain mode and a custom probe was employed for endoscopic detection of cervical cancer. Experiments for evaluating the measurement accuracy indicate that the model-to-data mismatch for the AC amplitude and phase lag is less than 3.7% and 6.7%, respectively. To facilitate the extraction of the optical properties in tubular tissues and to minimize the influence of the initial guess on the reconstruction accuracy, a fast inverse Monte Carlo simulation algorithm with cluster analysis method was proposed. Simulation results showed that the relative errors of the absorption coefficient recovered using the proposed inverse algorithm are less than 5.8% and those of the sacttering coefficient are less than 10.2%. Endoscopic measurement on two tubular solid phantoms were also carried out to evaluate the system and the inversion scheme. The results demonstrated that less than 20% relative error can be achieved.