Automated detection of retinal microaneurysms using integrated approach

Name: Ratanak Khoeun

ThaSH: Burapha University.Faculty of Science

ThaSH: Optic nerve

ThaSH: Retina

ThaSH: วิทยานิพนธ์ -- ภาษาอังกฤษ

Abstract: Microaneurysms appear as the first sign of diabetic retinopathy. Early detection can help to prevent vision lost from both non-proliferative diabetic retinopathy and proliferative diabetic retinopathy. Due to the fact that microaneurysms are small red lesions with variety of intensity, size, and shape, this leads to detection accuracy challenging. Most of the recent works in this domain used supervised learning techniques. Under these techniques, detecting microaneurysms candidates is the core component. Hence, our research proposes a novel approach to detect actual microaneurysms from digital fundus image. The proposed method consists of 4 main steps. In first step or preprocessing, Gaussian filter and contrast limited adaptive histogram equalization are applied to enhance the green plane of the RGB image. Next, in the second step, the four rotated matched filters are used to improve visualization of microaneurysm candidates and other red lesions. In the third step, area based noise and vessel removing, and Major and Minor Axis Length are applied. We use 89 images from public database DIARETDB1 as input. As the result, our proposed methodology provides the sensitivity, specificity, and accuracy of 84.94%, 95.8%, and 95.43% respectively

Burapha University. Library

Address: CHONBURI

Email: buulibrary@buu.ac.th

Name: Suwanna Rasmequan

Role: Principal advisor

Name: Annupan Rodtook

Role: Co-advisor

Created: 2017

Modified: 2565-02-22

Issued: 2019-07-04

วิทยานิพนธ์/Thesis

application/pdf

CallNumber: Th 612.843 Ra232A

eng

DegreeName: Master of Science

Level: Master's Degree

Descipline: Informatics

Grantor: Burapha University

©copyrights Burapha University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	58910018.pdf	37.92 MB

ใช้เวลา

0.034765 วินาที