Ghosting artifact reduction of polarization sensitive optical coherence tomography images through wavelet-FFT filtering

Undesirable cross-coupling between polarisation-maintaining (PM) fibers can result in detrimental ghost artefacts within polarisation sensitive optical coherence tomography (PS-OCT) images. Such artefacts combine with coherence noise stripes (originating from Fresnel reflections of optical components), complex-conjugate derived mirror-images and further irregular autocorrelation terms originating from the sample. Together, these artefacts can severely degrade the detected images, making quantitative measurements of the tissue birefringence challenging to perform. In this work, we utilize the recently presented wavelet-FFT filter1 to efficiently suppress these imaging artefacts entirely through post-processing. While the original algorithm was designed to suppress one-dimensional stripe artefacts, we extend this methodology to also facilitate removal of artefacts following a duplicate or inverse (mirror) profile to that of the skin surface. This process does not require any hardware modification of the system and can be applied retroactively to previously acquired OCT images. The performance of this methodology is evaluated by processing artefact-corrupted PS-OCT images of skin consisting of simultaneously detected horizontal and vertical polarized light. The resulting images are used to calculate a phase retardance map within the skin, the profile of which is indicative of localized birefringence. Artefacts in the resulting processed PSOCT images were notably attenuated compared to the unprocessed raw-data, with minimal degradation to the underlying phase retardation information. This should improve the reliability of curve-fitting for measurements of depth-resolved birefringence.