Techniques for Exudate Detection for Diabetic Retinopathy

Erwin, Erwin (2020) Techniques for Exudate Detection for Diabetic Retinopathy. In: 2019 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), 24-25 Oct 2019, Jakarta.

This is the latest version of this item.

[thumbnail of PID6134661-Erwin.pdf]
Preview
Text
PID6134661-Erwin.pdf - Accepted Version

Download (355kB) | Preview
[thumbnail of IEEE Xplore Search Results.pdf]
Preview
Text
IEEE Xplore Search Results.pdf

Download (122kB) | Preview

Abstract

Diabetic Retinopathy (DR) is an eye disease caused by diabetic complications that have early signs of a disease containing microaneurysms and exudates. Diabetic retinopathy for a long time can cause vision loss (blindness). So automatic detection is needed. Therefore we conduct research for the detection of exudates based on segmentation using the STARE and DIARETDB1 datasets. The exudate appears yellowish and glowing in the background of the retina with irregular size and shape. The use of several segmentation methods can be done in exudate detection. The method used is the adaptive threshold method, multi-threshold otsu, top-hat and bottom hat, and fuzzy c-means performance. The average performance results of several methods used in segmenting for each image in the STARE dataset are otsu multi threshold 87.1%, adaptived threshold of 89.9%, top-hat and bottom hat 87.7% and fuzzy cmeans 95.4%. and in the DIARETDB1 dataset, otsu multi threshold is 89%, adaptive threshold is 88.2%, top-hat and bottom hat 92.9% and fuzzy c-means 90.6%. These results indicate that the proposed method can provide good exudate segmentation results.

Item Type: Conference or Workshop Item (Paper)
Subjects: Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Divisions: 09-Faculty of Computer Science > 56201-Computer Systems (S1)
Depositing User: Dr Erwin Erwin
Date Deposited: 12 Oct 2020 00:43
Last Modified: 12 Oct 2020 00:43
URI: http://repository.unsri.ac.id/id/eprint/36523

Available Versions of this Item

Actions (login required)

View Item View Item