Elsevier

Knowledge-Based Systems

Volume 39, February 2013, Pages 9-22
Knowledge-Based Systems

Evolutionary algorithm based classifier parameter tuning for automatic diabetic retinopathy grading: A hybrid feature extraction approach

https://doi.org/10.1016/j.knosys.2012.09.008Get rights and content

Abstract

Human eye is one of the most sophisticated organ, with retina, pupil, iris cornea, lens and optic nerve. Automatic retinal image analysis is emerging as an important screening tool for early detection of eye diseases. Uncontrolled diabetes retinopathy (DR) and glaucoma may lead to blindness. DR is caused by damage to the small blood vessels of the retina in the posterior part of the eye of the diabetic patient. The main stages of DR are non-proliferate diabetes retinopathy (NPDR) and proliferate diabetes retinopathy (PDR). The retinal fundus photographs are widely used in the diagnosis and treatment of various eye diseases in clinics. It is also one of the main resources used for mass screening of DR. We present an automatic screening system for the detection of normal and DR stages (NPDR and PDR). The proposed systems involves processing of fundus images for extraction of abnormal signs, such as area of hard exudates, area of blood vessels, bifurcation points, texture and entropies. Our protocol uses total of 156 subjects consisting of two stages of DR and normal. In this work, we have fed thirteen statistically significant (p < 0.0001) features for Probabilistic Neural Network (PNN), Decision Tree (DT) C4.5, and Support Vector Machine (SVM) to select the best classifier. The best model parameter (σ) for which the PNN classifier performed best was identified using global optimization techniques such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). We demonstrated an average classification accuracy of 96.15%, sensitivity of 96.27% and specificity of 96.08% for σ = 0.0104 using threefold cross validation using PNN classifier. The computer-aided diagnosis (CAD) results were validated by comparing with expert ophthalmologists. The proposed automated system can aid clinicians to make a faster DR diagnosis during the mass screening of normal/DR images.

Introduction

Digital photography of the retina is widely used for screening of patients suffering from sight threatening diseases such as DR and glaucoma [11], [27], [3], [40]. Increase in the number of aging population, physical inactivity and obesity are contributing factors for diabetes. The global prevalence of diabetes is expected to rise from 2.8% in 2000 to 4.4% of the global population by 2030 [39]. Complication of diabetes leads to DR and a leading cause of blindness in the world. It is estimated that by the year 2010, the number of diabetic patients worldwide will be more than 221 million [66], [55].

Diabetes Retinopathy (DR) is a silent disease and may only be recognized by the patients when changes in the retina have progressed to a level, where the treatment becomes complicated or nearly impossible [72]. It can be broadly classified as NPDR and PDR depending on the presence of clinical features (microaneurysms, haemorrhages, hard exudates, cotton wool spots or venous loops) on the retina [14], [56]. A normal retina of the eye does not have any of the above cited features and is shown in Fig. 1a. In the NPDR stage (shown in Fig. 1b), the disease can advance from mild, moderate to severe stage with various levels of above said features except less growth of new blood vessels [56]. Fig. 1c is the typical PDR image, where the fluids sent by the retina for nourishment trigger the growth of new blood vessels. They grow along the retina and over the surface of the clear, vitreous gel that fills the inside of the eye. If they leak blood, may cause severe loss of vision and even result in blindness [56], [14], [73].

According to the US National Institute of Health (NIH) and National Eye Institute (NEI), the primary cause of the visual loss from DR is the failure to have regular eye examinations [72]. Proper screening followed by treatment by laser surgery can significantly reduce the incidence of blindness. It was believed that the routine screening and timely diagnosis can reduce the risk of blindness by 95% [72], [69]. Manual analysis and diagnosis require a great deal of time and energy to review the fundus images. Automated analysis and diagnosis would reduce the a large amount of time and effort [39]. In recent years, due to the steady growth of number of diabetic patients new tools and methodologies to facilitate the screening and evaluation procedures for DR were developed [55].

Many investigations have been carried out on the computer assisted analysis of the retinal fundus images [20], [22]. But these systems were not able to provide accurate investigations on different stages of DR. The automatic detection of microaneurysms, hard exudates, cotton wool spots, and haemorrhages for the pathology detection were studied [21], [30], [45]. These methods failed when applied to the large abnormal fundus images database. Many algorithms and techniques have been proposed to extract the features from the fundus images [46], [47].

The automatic detection of normal, mild DR, moderate DR, severe DR, and PDR stages using the bispectral invariant features of higher-order spectra techniques were proposed [1]. Further, the same authors proposed morphological features of fundus images for DR grading [2]. However, there is a need to improve diagnostic accuracy.

An algorithm to detect the optic disk (OD), blood vessels and fovea was developed [57], [19]. This algorithm worked effectively only for normal retinal images. Ege et al., have located the OD, fovea and four red and yellow abnormalities (microaneurysms, haemorrhages, exudates and cotton-wool-spots) in 38 color fundus images [10]. In this work the symptoms of abnormalities were graded by the ophthalmologist manually. An image analysis system for the automatic diagnosis of DR was developed [58]. The method developed in this work was not be able to detect large variations in the features of the abnormal retinal images. Algorithm for the computer based identification of microaneurysms was developed [61], but work did not provide any hope on the automatic identification of stages in DR and also lack in usage of large database. In recent years many researchers proposed system for the automatic identification of features for DR,to aid the automated diagnosis [28], [64], [74], [12].

The above discussed methods are mainly useful in analysis of the specific features on the retina, but do not provide a comprehensive system for the automatic detection of different stages of DR. The algorithms so far developed so far are unable to detect an early stage of retinopathy (NPDR) accurately. We are proposing a system for automated classification of normal, NPDR and PDR retinal images by automatically detecting the blood vessels, hard exudates, texture and entropy features. The proposed system is shown in block diagram is shown in Fig. 2. The features namely, blood vessels area, bifurcation (node) points in the blood vessels, exudates area, global texture and entropies are computed from the processed retinal images. These features were fed to the decision tree C4.5, SVM and Probabilistic Neural Network (PNN) classifiers to select the best classifier.

The layout of this paper is as follows: Section 2 deals with the data acquisition process, pre-processing, segmentation and extraction of features. Section 3 explains the classifiers used PNN, decision tree C4.5 and SVM. Section 4 presents the results of the system and discussion is covered in Section 5. Finally, paper concludes in Section 6.

Section snippets

Fundus image acquisition

The images used for this work were taken using TOPCON non-mydriatic retinal camera of model TRC-NW200. The built-in charge coupled device (CCD) camera provides up to 3.1 megapixels of high quality imaging. The inbuilt imaging software was used to store the images in the JPEG format. The data was acquired at Department of Ophthalmology, Kasturba Medical College, Manipal, India. The images were photographed and certified by the doctors in the department. The ethics committee consisting of senior

Diabetic retinopathy classification

In our work, we have used three classifiers namely support vector machine, decision tree C4.5 and PNN. They are briefly explained below.

Results

The features such as blood vessels area, exudates area, bifurcation point count, LBP energy, LBP entropy, Laws mask energy and entropies (Shannon, Kapur and Renyi) were extracted for three classes. We have extracted twenty-five features using above mentioned methods and only thirteen were found to be clinically significant (shown in Table 1]. The values of the blood vessel area and bifurcation point counts are increasing gradually from normal to PDR stage. It can also be seen from the table

Discussion

Several studies have been reported in the automated classification of DR images. In this section, we have summarized the classification of two, three, four and five classes of digital fundus images (normal and DR classes).

Conclusion

Diabetes retinopathy and glaucoma are the leading cause of blindness in the world. In this study, advanced image processing and machine learning algorithms were used to pre-process retinal digital images, extract the features and classify the retinal images to Normal, NPDR and PDR classes. Our experimental results show that the proposed method yields an average accuracy of 96.15%, sensitivity of 96.27%, and specificity of 96.08%. The higher accuracy obtained by the PNN classifier is due to the

Acknowledgment

Authors thank Social Innovation Research Fund (SIRF), Singapore for providing grant for this research.

References (75)

  • U.R. Acharya et al.

    Application of higher order spectra for the identification of diabetes retinopathy stages

    J. Med. Syst.

    (2008)
  • U.R. Acharya et al.

    Computer-based detection of diabetes retinopathy stages using digital fundus images

    Proc. Inst. Mech. Eng. H

    (2009)
  • U.R. Acharya et al.

    Automated diagnosis of glaucoma using texture and higher order spectra features

    IEEE Trans. Inform. Technol. Biomed.

    (2011)
  • A.M. Aibinu, M.I. Iqbal, M. Nilsson, M.J.E. Salami, Automatic Diagnosis of diabetic retinopathy from fundus images...
  • Z. Baochang et al.

    Local derivative pattern versus local binary pattern: face recognition with high-order local pattern descriptor

    IEEE Transactions on Image Processing

    (2010)
  • E.K.P. Chong et al.

    An Introduction to Optimization

    (2001)
  • N. Christianini et al.

    An Introduction to Support Vector Machines and Other Kernel based Learning Methods

    (2000)
  • M.J. Cree, J.J.G. Leandro, J.V.B. Soares, R.M.J.C. Jr, H.F. Jelinek, D. Cornforth, Comparison of various methods to...
  • B.M. Ege, O.V. Larsen, O.K. Hejlesen, Detection of abnormalities in retinal images using digital image analysis, in:...
  • Y.C. Emily

    Diabetic Retinopathy Preferred Practice Patterns

    (2003)
  • K.H. Englmeier et al.

    Early detection of diabetes retinopathy by new algorithms for automatic recognition of vascular changes

    Eur. J. Med. Res.

    (2004)
  • D.B. Fogel

    Evolutionary Computation: Toward a New Philosophy of Machine Intelligence

    (2006)
  • R.N. Frank

    Diabetic retinopathy

    Progr. Retinal Eye Res.

    (1995)
  • D. Goldberg

    Genetic Algorithms in Search, Optimization, and Machine Learning

    (1989)
  • R. Gonzalez et al.

    Digital Image Processing

    (2002)
  • A.M. Gun et al.

    An Outline of Statistical Theory

    (2005)
  • R. Gupta et al.

    The use of texture analysis to delineate suspicious masses in mammography

    Phys. Med. Biol.

    (1995)
  • W. Huan, H. Wynne, G. Kheng Guan, L. Mong Li, An effective approach to detect lesions in color retinal images, in:...
  • L. Huiqi, O. Chutatape, Automatic location of optic disk in retinal images, in: Proceedings of 2001 International...
  • A. Hunter, J. Lowell, J. Owens, L. Kennedy, Quantification of diabetic retinopathy using neural networks and...
  • M.I. Iqbal, A.M. Aibinu, M. Nilsson, I.B. Tijani, M. Salami, Detection of vascular intersection in retina fundus image...
  • H.J. Jelinek et al.

    An automated microaneurysm detector as a tool for identification of diabetic retinopathy in rural optometric practice

    Clin. Exp. Optom.

    (2006)
  • P. Kahai et al.

    A decision support framework for automated screening of diabetic retinopathy

    Int. J. Biomed. Imag.

    (2006)
  • J. Kennedy, R. Eberhart, Particle swarm optimization, in: Proceedings of the IEEE International Conference on Neural...
  • M.M.R. Krishnan, U.R. Acharya, Chua Kuang Chua, E.Y.K. Ng, L.C. Min, M.M. Mushrif, A. Laude, Application of...
  • M. Larsen et al.

    Automated detection of fundus photographic red lesions in diabetic retinopathy

    Invest. Ophthalmol. Vis. Sci.

    (2003)
  • N. Larsen et al.

    Automated detection of diabetic retinopathy in a fundus photographic screening population

    Invest. Ophthalmol. Vis. Sci.

    (2003)
  • Cited by (162)

    • Multimodal image encoding pre-training for diabetic retinopathy grading

      2022, Computers in Biology and Medicine
      Citation Excerpt :

      Firstly, this can be used as an intermediate step to facilitate the extraction of other anatomical structures or relevant lesions [7,34,37]. For instance, the localization or segmentation of the optic disc is typically used as an intermediate step in the detection of bright lesions such as exudates [34], whereas the segmentation of the blood vessels is useful for the detection of red lesions such as hemorrhages or microaneurysms [37]. Secondly, some anatomical structures can be directly used for the extraction of anatomical features that are relevant for the diagnosis of DR, such as vessel areas [33,35] or vessel bifurcation points [34,35].

    • Optical coherence tomography in diabetic retinopathy: A review in application of artificial intelligence

      2023, Photo Acoustic and Optical Coherence Tomography Imaging, Volume 1: Diabetic retinopathy
    • Intelligent analysis of fundus images: Methods and applications

      2023, Intelligent Analysis Of Fundus Images: Methods And Applications
    View all citing articles on Scopus
    View full text