Retinal Image Classification for Diagnosis of Vogt-Koyanagi-Harada Disease Open Access

Feng, Haonan (2017)

Permanent URL:


Vogt-Koyanagi-Harada (VKH) disease is a disease with four stages and each stage has different symptoms. The accurate identification of different stage is of increasing importance in diagnosis and treatment of VKH disease. In this work, Sunset-Glow Fundus (SGF) appearance and Serous Retinal Detachments (SRD), two of the most significant symptoms shown in the retinal images when VKH disease develops, are chosen as the feature to identify the stage of VKH disease. To extract the information of these features precisely, two image preprocessing methods are applied at the beginning. In the feature extraction section, the distribution of color intensity, Fast Fourier Transformation, and the Grey-Level Co-occurrence method will be introduced to construct the feature matrix for model training. A k-nearest neighbor algorithm and an optimized support vector machine are employed as the classifiers. A sample dataset consisting of 144 retinal images is used for identifying four different stage of VKH disease. With optimized model combined with two SVM classifiers, the classification accuracy of 86.8% will be obtained.

Table of Contents

1 Introduction

2 Method

2.1 Image Preprocessing

2.2 Feature Extraction

2.2.1 Color Distribution

2.2.2 Fast Fourier Transformation

2.2.3 Grey-Level Co-occurence Matrix

2.3 Model Training

3 Result

4 Conclusion and Discussion

5 Reference

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Last modified

Primary PDF

Supplemental Files