Automated Microaneurysm Detection in Fundus Images by Region Growing

Lin Li • Juan Shan

09:05 - 09:55 | Thursday 16 February 2017 | Ballroom D


In this study, we propose an approach to automatically partition a fundus image into regions and evaluate whether a region contains microaneurysm (MA). The approach can achieve sensitivity 81%, specificity 95.4%, and accuracy 91.8%, on a public dataset DIARETDB.