Oxygen-induced retinopathy (OIR) is a widely used model to study ischemia-driven neovascularization (NV) in the retina and to serve in proof-of-concept studies in evaluating antiangiogenic drugs for ocular, as well as nonocular, diseases. The primary parameters that are analyzed in this mouse model include the percentage of retina with vaso-obliteration (VO) and NV areas. However, quantification of these two key variables comes with a great challenge due to the requirement of human experts to read the images. Human readers are costly, time-consuming, and subject to bias. Using recent advances in machine learning and computer vision, we trained deep learning neural networks using over a thousand segmentations to fully automate segmentation in OIR images. While determining the percentage area of VO, our algorithm achieved a similar range of correlation coefficients to that of expert inter-human correlation coefficients. In addition, our algorithm achieved a higher range of correlation coefficients compared with inter-expert correlation coefficients for quantification of the percentage area of neovascular tufts. In summary, we have created an open-source, fully automated pipeline for the quantification of key values of OIR images using deep learning neural networks.
Sa Xiao, Felicitas Bucher, Yue Wu, Ariel Rokem, Cecilia S. Lee, Kyle V. Marra, Regis Fallon, Sophia Diaz-Aguilar, Edith Aguilar, Martin Friedlander, Aaron Y. Lee
Guidelines: The Editorial Board will only consider letters that we deem relevant and of interest to our readers. We will not post data that have not been subjected to peer review, nor will we post letters that are essentially a reiteration of another letter. All accepted letters will be posted on our website within one week of acceptance. We reserve the right to edit any letter for length, content, and clarity. Authors of all accepted letters will be asked to preview any changes. Authors will be notified by e-mail if their letters were not accepted. As this is a final decision, no appeals will be considered.
Specific requirements: All letters must be 400 words or fewer. You may enter the letter as plain text or HTML. The author's name and e-mail address are required, and will be posted with the letter. All possible conflicts of interest must be noted, even if they are not posted. If you wish to include a figure (keep in mind that non-peer-reviewed data will not be posted), please contact the editors directly at firstname.lastname@example.org.