Differential autocorrelation histogram of singular values for image reconstruction in reflection matrix optical coherence tomography.

Abstract

Reflection matrix optical coherence tomography has recently emerged as a promising modality for ultra-deep optical imaging in a turbid tissue. However, a critical issue that is not well addressed is to determine the optimized number of primary singular values of the measured reflection matrix used for image reconstruction. In this paper, the method based on the differential autocorrelation histogram of singular values is proposed. The histogram distribution generally consists of two regions with distinct distribution characteristics. One is the continuously distributed region, which is related to the multiple-scattered noise and can be fitted by a skewed distribution curve. The other is the discrete distribution region associated with the single-scattered signal, which does not obey the fitting curve. The boundary between two regions is then identified and used to determine the optimized number of the primary singular values for image reconstruction. Both simulating and experimental results demonstrate that the proposed method is more robust and adaptable to samples with different signal-to-noises and large dynamic ranges of reflectivity, outperforming two currently adopted methods.