Numerical Study of Supra-Wavelength Axial Motion Compensation in Contact-Mode Optical Coherence Angiography Using Fourier-Shift Procedures

Abstract

Label-free angiographic methods based on optical coherence tomography (OCT) visualize blood vessels utilizing detection of red blood cells motion against surrounding static tissue. However, in practice, the surrounding tissue is never still due to natural motions of living organisms (e.g., breathing or heart beating). To mitigate large scale motions of the tissue relatively to the OCT probe the tissue examination can be made in contact mode. In such a case, however, the OCT probe inevitably exerts some pressure onto the tissue, so that bulk motions lead to interframe deformations and depth-dependent tissue displacements which have to be numerically compensated prior to angiographic visualization. Usually, sufficiently small deformations primarily affect pixel phases in OCT images rather than pixel amplitudes, and, therefore, phase-only compensation of the masking motions may be fairly sufficient. However, in case of larger strains and supra-wavelength displacements, larger inter-scan phase variations of the order of several periods lead to the appearance of pronounced “decorrelation noise” in which variations in pixel amplitudes and phases are combined. This effect significantly degrades the quality of the final OCT-angiography images. In this paper, we present a new method allowing to a significant degree to compensate this phase-amplitude decorrelation caused by spatially-inhomogeneous supra-wavelengths displacements. This compensation is based on the Fourier-shift theorem which allows one to back-shift fragments of the deformed OCT-scans to their initial positions before deformation. At the same time variations of pixels due to the motion of blood particles within smaller-in-size vessel cross sections are retained. Although such backshifts do not compensate relative motions of sub-resolution particles, this procedure efficiently reduces decorrelation even for fairly big spatially-inhomogeneous displacements and leads to much lower signal variability outside blood vessels while preserving high variability inside. The proposed compensation method is compared to the earlier proposed phase-only compensation using simulated data. Pronouncedly lower strain-induced artefacts and much higher contrast between blood vessels and background are demonstrated.