抄録
For Optical Coherence Tomography (OCT) images we propose a new automatic extraction and segmentation of retinal layer structures for assisting ophthalmologists in clinical decision such as diagnosis and treatment. The previous automatic method (One Directional Active Grid; ODAG)) proposed by our group can extract only extracts Inner Limiting Membrane (ILM) and Retinal Pigment Epithelium (RPE). In this paper, we propose a new multi-boundary extraction method, that is, two-step One Dimensional Active Grid (two-step ODAG). The first step is the previous ODAG, and the second step is a new process to extract multi-boundary in retinal layer. We set the initial position of nodes between ILM and RPE, and we are developing a new operation window to detect multi-boundary in the retinal region. We implemented our proposed system by C language and OpenCV library on MS-Windows. We evaluated the accuracy for 10 normal OCT images, and the accuracy is 90%.
