3D Automatic Segmentation Method for Retinal Optical Coherence Tomography Volume Data Using Boundary Surface Enhancement

摘要

With the introduction of spectral-domain optical coherence tomography(SDOCT), much larger image datasets are routinely acquired compared to what waspossible using the previous generation of time-domain OCT. Thus, there is acritical need for the development of 3D segmentation methods for processingthese data. We present here a novel 3D automatic segmentation method forretinal OCT volume data. Briefly, to segment a boundary surface, two OCT volumedatasets are obtained by using a 3D smoothing filter and a 3D differentialfilter. Their linear combination is then calculated to generate new volume datawith an enhanced boundary surface, where pixel intensity, boundary positioninformation, and intensity changes on both sides of the boundary surface areused simultaneously. Next, preliminary discrete boundary points are detectedfrom the A-Scans of the volume data. Finally, surface smoothness constraintsand a dynamic threshold are applied to obtain a smoothed boundary surface bycorrecting a small number of error points. Our method can extract retinal layerboundary surfaces sequentially with a decreasing search region of volume data.We performed automatic segmentation on eight human OCT volume datasets acquiredfrom a commercial Spectralis OCT system, where each volume of data consisted of97 OCT images with a resolution of 496 512; experimental results show that thismethod can accurately segment seven layer boundary surfaces in normal as wellas some abnormal eyes.