Denoise method for Reflection Matrix Optical Coherence Tomography

Reflection matrix optical coherence tomography (RM-OCT) has recently emerged as a solution to address the limited imaging depth in conventional optical coherence tomography (OCT). The RM-OCT image relies on the reflection matrix reconstruction technique, assuming distinct thresholds between multiply scattered and singly scattered photons. By setting this threshold, the optical field information composed of singly scattered photons is reconstructed. However, a significant challenge arises when imaging strong reflection regions, where the singly scattered photons from strong reflections contribute equally to those from the target, causing the imaging target to be submerged in strong reflection backgrounds. This study proposes a novel noise reduction technique based on robust principal component analysis (RPCA), applied for the first time in RM-OCT image. In Monte Carlo experiments, images were successfully decomposed into a low -rank matrix representing background noise and a sparse matrix representing the imaging target. Further experiments demonstrated a significant increase of 7.79 dB in the signal-to-noise ratio of the target image, greatly enhancing imaging quality and accuracy. Biological tissue imaging experiments confirmed the unique advantage of this method, which not only reduces background noise but also highlights imaging target details.