Beyond Retinal Layers: An Automatic Active Contour Model with Pre-Fitting Energy for Subretinal Fluid Segmentation in SD-OCT Images

Abstract

Automatically and accurately segment neurosensory retinal detachment (NRD) associated subretinal fluid in spectral-domain optical coherence tomography (SD-OCT) is vital for the evaluation of central serous chorioretinopathy (CSC). A two-stage unsupervised fluid segmentation algorithm is proposed. In the first stage, the candidate fluid region is automatically estimated to obtain the initial curve of the fluid area for the level set method. In the second stage, the local Gaussian pre-fitting energy model is proposed to segment subretinal fluid. The testing data set with 23 longitudinal SD-OCT cube scans from 12 eyes of 12 patients are used to evaluate the proposed algorithm. Comparing with two independent experts' manual segmentations, our algorithm obtained a mean positive predicative value 94.0% and dice similarity coefficient 94.4%, respectively. Without retinal layer segmentation, the proposed algorithm can obtain high segmentation accuracy. Our model may provide reliable subretinal fluid segmentations for NRD from SD-OCT images and shows the potential to improve clinical therapy for CSC.