DISCOVER: 2-D multiview summarization of Optical Coherence Tomography Angiography for automatic diabetic retinopathy diagnosis

Mostafa El Habib Daho, Yihao Li, Rachid Zeghlache, Hugo Le Boité, Pierre Deman, Laurent Borderie, Hugang Ren, Niranchana Manivannan, Capucine Lepicard, Béatrice Cochener, Aude Couturier, Ramin Tadayoni, Pierre-Henri Conze, Mathieu Lamard, Gwenolé Quellec

Abstract: Highlights•An algorithm to quickly and efficiently classify 3-D images is presented.•3-D images are summarized by one or a few 2-D images as an intermediate step.•The pipeline is trained from end to end: the 2-D summary is thus problem-specific.•This multiview 2-D summarization allows interpretability and knowledge discovery.•It is applied to diabetic retinopathy severity assessment using OCT angiography.