Noise Reduction of Swept-Source Optical Coherence Tomography via Compressed Sensing

Abstract

In this paper, we investigate noise reduction in swept-source optical coherence tomography (OCT) using compressed sensing (CS). Multiple scan averaging is a classical method used to enhance the quality of OCT images by reducing the noise of a system. However, the conventional averaging method requires a repetitive scan at the same location and thus reduces the imaging speed. In this paper, the sparsity property of an OCT A-scan is utilized, and one full A-scan OCT image can be reconstructed from a portion of the acquired data during one sweep period using CS. Thus, multiple OCT A-scans can be reconstructed from a single sweep. The average A-scans yield a better quality than the single A-scan obtained from the whole data acquired during a sweep period. We demonstrate that the average of five reconstructed A-scans from a single sweep using CS offers an image quality and depth resolution similar to those obtained by averaging three sequential A-scans from three sweeps using the conventional averaging method. This proposed method can shorten the time required to perform repetitive scans and thus improve the imaging speed.