Drusen detection based on Scale-space with Feature Stability

Duanggate, C, Uyyanonvara, B, Makhanov, S, Barman, S A and Williamson, T (2011) Drusen detection based on Scale-space with Feature Stability. In: Medical Image Analysis and Understanding MIUA 2011; 14-15 Jul 2011, London, United Kingdom.


This paper proposes a novel segmentation technique for drusen detection in retinal fundus images based on the scale space approach endowed with feature stability. To select significant blobs representing drusen, the stability of the features of scale-space blobs is taken into account in addition to conventional blobs' lifetime. The algorithm was tested with over 20,000 blobs from 26 retinal images and the results show that the method can detect variable size and variable shape drusen efficiently with Positive Predictive Value over 95%.

Actions (Repository Editors)

Item Control Page Item Control Page