作者: Menglin Wu , Wen Fan , Qiang Chen , Zhenlong Du , Xiaoli Li
DOI: 10.1364/BOE.8.004257
关键词:
摘要: Assessment of serous retinal detachment plays an important role in the diagnosis central chorioretinopathy (CSC). In this paper, we propose automatic, three-dimensional segmentation method to detect both neurosensory (NRD) and pigment epithelial (PED) spectral domain optical coherence tomography (SD-OCT) images. The proposed involves constructing a probability map from training samples using random forest classification. is constructed linear combination structural texture, intensity, layer thickness information. Then, continuous max flow optimization algorithm applied segment detachment-associated fluid regions. Experimental results 37 SD-OCT volumes cases CSC demonstrate can achieve true positive volume fraction (TPVF), false (FPVF), predicative value (PPV), dice similarity coefficient (DSC) 92.1%, 0.53%, 94.7%, 93.3%, respectively, for NRD 92.5%, 0.14%, 80.9%, 84.6%, PED segmentation. be automatic tool evaluate has potential improve clinical evaluation CSC.