Artificial neural network (ANN) for dispersion compensation of spectral domain – optical coherence tomography (SD-OCT)

Abstract

Abstract Dispersion is a factor that causes the axial resolution to decrease in optical coherence tomography (OCT). In this paper, an artificial neural network (ANN) is reported for correcting the dispersion problem. By training a neural network, the dispersion-free spectral signal is predicted only from a given interference spectral signal. First, the dispersion principle of OCT is analyzed. Next, the process for finding the global optimum of dispersion-free spectral signal distribution is described as a training process, and the ANN model is introduced. Lastly, simulation and experiments show that the presented method improves the axial resolution of the system. Accordingly, the ANN model fits the non-linear relationship between input and output, and the spectral response shows the problem of full width at half maximum (FWHM) due to dispersion in OCT which is of great significance.