Early Detection On Diabetic Retinopathy Stage Level Based On Knowledge Base Using Extreme Learning Machine Method

Fitriati, Desti (2015) Early Detection On Diabetic Retinopathy Stage Level Based On Knowledge Base Using Extreme Learning Machine Method. ICo-ApICT 2015.

Full text not available from this repository.
Official URL: http://elib.unikom.ac.id/gdl.php?mod=browse&op=rea...

Abstract

Diabetic retinopathy (DR) is a disease that can lead to blindness. It occurs as a result of microvascular damage in retina and also influenced by diabetes mellitus. It��s like a nightmare because it can even affect the quality of life in social, emotional, and psychological. Blindness process in DR through several stages, start from the early stage to the advanced stage. Therefore the early detection is very important to know the appropriate prevention and treatment. At this research the stage level is divided into Normal, Early NPDR, Advanced NPDR and PDR. The stage classification based on the information of retina data taken from Cipto Mangunkusumo Hospital (RSCM) in Jakarta. There are two mechanism in this experiment, the first uses texture feature of Grey Level Co-occurrence Matrix (GLCM), and the second uses the unique DR features such as microaneurysm, exudates, and blood vessels. The training and testing data is divided using 10-fold Cross Validation. Then the data is processed using morphological techniques and classified using the Extreme Learning Machine (ELM) which known 1000 times faster than the Single Layer Perceptron and it is also suitable for medical image learning. This research resulted the basic information that can be used as a reference of DR stage level category with the accuracy 53.53% for the first mechanism and 52.26% for the second mechanism.

Item Type: Article
Uncontrolled Keywords: Diabetic Retinopathy; Cross Validation; GLCM, Morfologi; Extreme Learning Machine
Subjects: Prosiding > ICo-ApICT 2015
Divisions: Universitas Komputer Indonesia > Perpustakaan UNIKOM
Date Deposited: 10 Dec 2015 10:25
Last Modified: 31 Jan 2019 10:25
URI: https://repository.unikom.ac.id/id/eprint/59795

Actions (login required)

View Item View Item