DREAM: Diabetic Retinopathy Analysis Using Machine Learning

被引:214
作者
Roychowdhury, Sohini [1 ]
Koozekanani, Dara D. [2 ]
Parhi, Keshab K. [1 ]
机构
[1] Dept Elect & Comp Engn, Minneapolis, MN 55455 USA
[2] Univ Minnesota, Dept Ophthalmol & Visual Neurosci, Minneapolis, MN 55455 USA
关键词
Bright lesions; classification; diabetic retinopathy (DR); fundus image processing; red lesions; segmentation; severity grade; AUTOMATIC DETECTION; RETINAL IMAGES; SYSTEM; MICROANEURYSMS; CLASSIFICATION; DIAGNOSIS;
D O I
10.1109/JBHI.2013.2294635
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
This paper presents a computer-aided screening system (DREAM) that analyzes fundus images with varying illumination and fields of view, and generates a severity grade for diabetic retinopathy (DR) using machine learning. Classifiers such as the Gaussian Mixture model (GMM), k-nearest neighbor (kNN), support vector machine (SVM), and AdaBoost are analyzed for classifying retinopathy lesions from nonlesions. GMM and kNN classifiers are found to be the best classifiers for bright and red lesion classification, respectively. A main contribution of this paper is the reduction in the number of features used for lesion classification by feature ranking using Adaboost where 30 top features are selected out of 78. A novel two-step hierarchical classification approach is proposed where the nonlesions or false positives are rejected in the first step. In the second step, the bright lesions are classified as hard exudates and cotton wool spots, and the red lesions are classified as hemorrhages and micro-aneurysms. This lesion classification problem deals with unbalanced datasets and SVM or combination classifiers derived from SVM using the Dempster-Shafer theory are found to incur more classification error than the GMM and kNN classifiers due to the data imbalance. The DR severity grading system is tested on 1200 images from the publicly available MESSIDOR dataset. The DREAM system achieves 100% sensitivity, 53.16% specificity, and 0.904 AUC, compared to the best reported 96% sensitivity, 51% specificity, and 0.875 AUC, for classifying images as with or without DR. The feature reduction further reduces the average computation time for DR severity per image from 59.54 to 3.46 s.
引用
收藏
页码:1717 / 1728
页数:12
相关论文
共 39 条
[11]   Robust Detection of Microaneurysms for Sight Threatening Retinopathy Screening [J].
Bhalerao, Abhir ;
Patanaik, Amiya ;
Anand, Sarabjot ;
Saravanan, Pounnusamy .
SIXTH INDIAN CONFERENCE ON COMPUTER VISION, GRAPHICS & IMAGE PROCESSING ICVGIP 2008, 2008, :520-+
[12]   LIBSVM: A Library for Support Vector Machines [J].
Chang, Chih-Chung ;
Lin, Chih-Jen .
ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2011, 2 (03)
[13]  
Cherkassky V, 2007, LEARNING DATA CONCEP
[14]  
Cuadros J., 2009, J DIABETES SCI TECHN, V3
[15]  
Esnaashari Monireh, 2011, International Journal of Research and Reviews in Computer Science, V2, P1222
[16]   A comparison of computer based classification methods applied to the detection of microaneurysms in ophthalmic fluorescein angiograms [J].
Frame, AJ ;
Undrill, PE ;
Cree, MJ ;
Olson, JA ;
McHardy, KC ;
Sharp, PF ;
Forrester, JV .
COMPUTERS IN BIOLOGY AND MEDICINE, 1998, 28 (03) :225-238
[17]  
Garg S., 2009, CLIN DIABETES, V27, P140
[18]  
Kauppi T, 2007, DIARETDB1 DIABETIC R
[19]  
Lawrence Mary Gilbert, 2004, Trans Am Ophthalmol Soc, V102, P321
[20]   Automatic detection of red lesions in digital color fundus photographs [J].
Niemeijer, M ;
van Ginneken, B ;
Staal, J ;
Suttorp-Schulten, MSA ;
Abràmoff, MD .
IEEE TRANSACTIONS ON MEDICAL IMAGING, 2005, 24 (05) :584-592