Two papers accepted for OMIA 2019!
Our papers in OCT and fluorescein angiography segmentation were accepted for presentation at the MICCAI 2019 International Workshop in Ophthalmic Image Analysis (OMIA) in Shenzhen, China.
We got two papers accepted for publication at the OMIA workshop in MICCAI 2019 in Shengzen, China.
In “An amplified-target loss approach for photoreceptor layer segmentation in pathological OCT scans”, Anna Breger and I introduced an approach to increase penalization of errors in the central area of the input B-scan. By means of this approach, we force the neural network to take more into account the abnormalities that appear in the center of the scans in pathological retinas, improving results in photoreceptor layer segmentation. My other co-authors are Hrvoje Bogunović, Sophie Riedl, Bianca S. Gerendas, Martin Ehler and Ursula Schmidt-Erfurth.
In “Foveal Avascular Zone Segmentation in Clinical Routine Fluorescein Angiographies Using Multitask Learning”, Dominik Hofer introduces a multitask learning approach for FAZ segmentation in FA images. To the classical segmentation branch, we add a second decoder that outputs an Euclidean distance map. By penalizing errors both in segmentation and in this regression map, we increase the regularization of the model and improve its results in clinical routine images. The other co-authors are Philipp Seeböck, Georgios Mylonas, Felix Goldbach, Amir Sadeghipour, Bianca S Gerendas, Ursula Schmidt-Erfurth. This is part of Dominik’s PhD thesis, supervised by Bianca S Gerendas and Philipp Seeböck.
José Ignacio Orlando
Associate Researcher
My research interests include machine learning and computer vision techniques for medical imaging applications, mostly centered in ophthalmology.