Correction of motion artifact in 3D retinal optical coherence tomography imaging

In the system of spectral domain optical coherence tomography(SD-OCT), the location accuracy of scanning mechanism, mechanical dither and unconscious movement of human eyes can lead to the dither of cross-sectional scanning signal (B-scans) and affect the imaging quality of eyes. In this paper, we firstly make space transformation and filtering towards to the original retina images. Secondly, we find out the characteristic curve, which is called inner segment/outer segment (IS/OS) layer, and its location based on mathematical morphology opening and closing operation segmentation image and canny algorithm. Then we use the method of smoothing spline curve fitting to make correction towards to the characteristic curve and evaluate the fitting curve based on the structure similarity algorithm. Finally, we correct the 3D image according to the relative location of fitting curve and characteristic curve. The correction result of OCT image artifact in eyes bitterly throw away the artifact causing by mechanical dither and eyes movement, which lays the basis for adjuvant therapy of eyes diseases in the future.