Output details
11 - Computer Science and Informatics
University of Dundee
Accurate estimation of retinal vessel width using bagged decision trees and an extended multiresolution Hermite model
<23> Novel algorithm estimating vessel calibre in retinal images using machine learning. Thorough experiments show improvement on state-of-the-art. Collaborative with the Univ of Palermo (Italy). Resulting from a 2-year research, this algorithm is the current calibre estimation tool in our VAMPIRE package, used in clinical and biomarker studies in UK, US, Europe. First ever software applied to UK Biobank retinal images, in Edinburgh-Dundee collaborative project (2013); basis for CARMEN project vessel measument (2012-2015) with OPTOS plc and Univ of Edinburgh. Work propitiated invitations (Trucco) as plenary speaker to IEEE ISSNIP 2013 (Brazil) and LACNEM 2013 (Colombia).