Data for: Retinal layer segmentation in rodent OCT images: Local intensity profiles & fully convolutional neural networks
Database of rodent OCTs composed of two batches of images. Animal experiment permission was granted by the Danish Animal Experimentation Council (license number: 2017-15-0201-01213). Rat OCT images were taken with the Micron IV equipment (Phoenix Research Labs, Pleasanton, USA). All rats were anesthetized previously to image acquisition. A set of 22 Sprague-Dawley rats were used. The first batch is composed of 129 OCT images and the second one of 115 images (image size: 1024x1024 pixels; resolution: 0.9775 um/pixel). The most significant retinal layer boundaries that were visibly distinguishable on these images were segmented: ILM (Internal Limiting Membrane), IPL-INL (Inner Plexiform Layer - Inner Nuclear Layer), INL-OPL (Inner Nuclear Layer - Outer Plexiform Layer), OPL-ONL (Outer Plexiform Layer - Outer Nuclear Layer), IS-OS (Inner Segment - Outer Segment), RPE (Retinal Pigment Epithelium). The database contains the original rodent oct images obtained from circular scans, expert-reviewed segmentation (ground truth) and several automatic segmentation results for comparison purposes. This database must be loaded with MATLAB. For more information, see Sandra Morales, Adrián Colomer, José M. Mossi, Rocío del Amor, David Woldbye, Kristian Klemp, Michael Larsen, Valery Naranjo, Retinal layer segmentation in rodent OCT images: Local intensity profiles & fully convolutional neural networks, Computer Methods and Programs in Biomedicine, Volume 198, 2021, 105788, ISSN 0169-2607, https://doi.org/10.1016/j.cmpb.2020.105788.