Speaker "Roberto Reif" Details Back



Classification of diabetic and non-diabetic retinopathy from optical coherence tomography images using transfer learning


The goal of this project is to diagnose diabetic retinopathy from Optical Coherence Tomography images obtained from the retina of the eye.  Specifically, we used a Deep Learning model, such as Convolutional Neural Networks (CNN), to perform the image classification.  Given that we have a small dataset, we use a transfer learning technique.


I am currently a Senior Data Scientist at Metis and the founder of Analysis Consulting LLC, a data science and data visualization firm. I am a scientist with a strong background in data analysis, and image/signal processing. I have worked on applications in the healthcare, internet of things and business intelligence market. Previously I studied electrical and biomedical engineering, and have work experience in large companies, startups, and academic/research institutions.