Accelerated phase aberration compensation using the fast principal component analysis algorithm

Abstract

An accelerated phase aberration compensation method is proposed. The method uses the computationally efficient fast principal component analysis algorithm (FPCA) instead of the singular value decomposition (SVD) based PCA. FPCA bypasses the computation of all principal components aside from the first principal component (PC1), which is what is used to approximate the phase aberration. The removal of the phase aberration term was demonstrated on a digitally reconstructed wavefront of a cheek cell. It was found that phase compensation using FPCA was not only 10x faster compared to PCA by SVD but also takes up lesser storage space. It was also shown that the efficiency of FPCA compared to PCA by SVD increases for images with higher dimensionality. The proposed technique has the advantage of preserving the whole spectrum of the image without trading off an increase in computational time. The proposed technique is visioned to be helpful in realizing high-speed quantitative phase imaging methods with automatic phase compensation.