Early Diabetic Retinopathy (DR) identification through systematic screening with timely treatment are important steps to prevent blindness. In this sense, several researchers have focused their work on the development of different computer-aided lesion-specific detectors associated with DR. However, combining different detectors is a complex task since frequently the detectors have different properties and constraints and are not designed under a unified framework. In this paper, we extend upon our previous work for detecting DR lesions based on points of interest and visual words. The novelty of our work is in the creation of new detectors for several of the most common DR lesions and also in the investigation of fusion techniques aiming at combining different classifications toward a unified final answer whether an image for diagnosis is normal or presents signs of diabetic retinopathy. The combination methods show promising results and shed light on the possible advantages of combining complementary lesion detectors for the DR diagnosis problem.
@inproceedings{jelinek2012data,
authors = "Hebert F. Jelinek and Ramon Pires and Rafael Padilha and Siome Goldenstein and Jacques Wainer and Terry Bossomaier and Anderson Rocha",
title = "Data fusion for multi-lesion diabetic retinopathy detection",
booktitle = "IEEE International Symposium on Computer-Based Medical Systems (CBMS)",
year = 2012,
address = "Rome, Italy",
}