Adaptive filtering of optical coherent tomography fringe data with ensemble empirical mode decomposition

Empirical mode decomposition (EMD) is a new adaptive data analysis method in which the analyzed data is decomposed into a limited number of intrinsic mode functions (IMFs) through a sifting process. One problem with EMD is mode mixing, which has been solved by Wu et al using ensemble EMD (EEMD). In this paper, we applied the EEMD method to data acquired from optical coherence tomography (OCT) to improve the image quality. First, the original OCT fringe data is converted from linear wavelength to linear frequency through a calibration process. Second, the calibrated data is decomposed into different IMFs by EEMD. Third, the physical meaning of different IMFs was analyzed. Fourth, IMFs that represented noise were removed from the calibrated fringe data. The noise removed fringe data was then Fourier transformed to get depth information. EEMD was found to be able to separate different frequency noise into different IMFs. The signal to noise ratio of OCT image was improved by removing the IMFs that represent noise from the acquired fringe data.