研究目的
To segment and detect three types of fluid including sub-retinal fluid (SRF), intra-retinal fluid (IRF) and pigment epithelium detachment (PED) in OCT Bscans of subjects with age-related macular degeneration (AMD) and retinal vein occlusion (RVO) or diabetic retinopathy using a fully-automated method based on graph shortest path algorithms and convolutional neural network (CNN).
研究成果
The proposed fully-automated method effectively segments and detects three types of fluid in OCT volumes from subjects with AMD and RVO, achieving high Dice coefficients across different manufacturers' datasets. The method's effectiveness was also demonstrated in the 2017 Retouch challenge.
研究不足
The method's performance is lower in the Cirrus dataset with low signal-to-noise ratio. Future work includes fine-tuning the CNN for better results and addressing reproducibility studies between segmentation following repeat imaging.
1：Experimental Design and Method Selection:
The method involves segmenting inner limiting membrane (ILM) and retinal pigment epithelium (RPE) layers using graph shortest path methods, then inputting the regions between these layers to a CNN for binary classification of pixels between ILM and RPE. PED is segmented by layer flattening.
2：Sample Selection and Data Sources:
The dataset includes 112 macula-centered OCT volumes from the RETOUCH fluid segmentation challenge.
3：List of Experimental Equipment and Materials:
OCT volumes were acquired with spectral-domain SD-OCT devices from three different vendors: Cirrus HD-OCT (Zeiss Meditec), Spectralis (Heidelberg Engineering), and T-1000/T-2000 (Topcon).
4：Experimental Procedures and Operational Workflow:
The process includes layer segmentation, PED segmentation with layer flattening, IRF and SRF segmentation with CNN, and fluid detection.
5：Data Analysis Methods:
The performance is evaluated using Dice coefficients for IRF, SRF, and PED segmentation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
