Date of Award

5-2019

Document Type

Campus Access Thesis

Degree Name

Master of Science (MS)

Department

Computer Science

First Advisor

Marc Pomplun

Second Advisor

Dan Simovici

Third Advisor

Elizabeth O'Neil

Abstract

Diabetic Retinopathy (DR) is a leading cause of vision loss among people with diabetes and is mainly caused by damages to blood vessels in the retinal region of the eye. However, the good news is that it is preventable in 98% of the patients provided the symptoms are diagnosed at an early stage. What makes diagnosis challenging is that there are about 415 million people in the world with diabetes and over 30% of them are at risk of developing DR related vision complications. The current ophthalmology screening process is both expensive and time consuming, making it infeasible to screen all diabetes patients for DR. A key step in screening involves identifying visible deviations in the retinal images as compared to the images from a healthy individual's eye. There are clinically documented patterns like circular red spots, swelling of tissues, yellow material deposits and unusual blood vessels that are key to identifying occurrence of DR. This thesis presents Deep Learning (DL) models constructed using Google's Inception framework for recognizing these unusual patterns in retinal images. The performance of these models, image preprocessing techniques and the benefits of using Google Cloud Platform (GCP) are also discussed.

Comments

Free and open access to this Campus Access Thesis is made available to the UMass Boston community by ScholarWorks at UMass Boston. Those not on campus and those without a UMass Boston campus username and password may gain access to this thesis through resources like Proquest Dissertations & Theses Global or through Interlibrary Loan. If you have a UMass Boston campus username and password and would like to download this work from off-campus, click on the "Off-Campus UMass Boston Users" link above.

Download
Off-Campus UMass Boston Users

Share

COinS