Automatic Diagnosis of Glaucoma Using Digital Fundus Images
6.5. Conclusion
In this chapter, a simple novel method for the automatic diagnosis of glaucomatous abnormal eye through fundus images was developed using imageprocessing techniques and GMM. This proposed system can identify the unknown image with an accuracy of more than 83%. A GUI was developed to show the results and progression of the features. This system can also be used by physicians who would like to cross-check their glaucoma diagnosis.
References
1.Hitchings, R.A. and Spaeth, G.L. The optic disc in glaucoma, ii: Correlation of appearance of the optic disc with the visual field. Br J Ophthalmol 61:107–113, 1977.
2.Bulletin of World Health Organization, http://www.who.int/bulletin/volumes/82/11/ resnikoff1104abstract/en/.
3.G.R. Foundation, Primary open angle glaucoma, http://www.glaucoma.org/learn/types. php.
4.Nayak, J., Bhat, P.S., Acharya, U.R., Lim, C.M., and Kagathi, M. Automated identification of different stages of diabetic retinopathy using digital fundus images. J Med Syst 32:107–115, 2008.
5.Heiting, G., Haddrill, M., and Slonim, C. Narrow-angle glaucoma, http://www. allaboutvision.com/conditions/narrow-angle-glaucoma.htm.
6.Acharya, U.R., Chua, K.C., Ng, E.Y.K., Wei, W., and Chee, C. Application of higher order spectra for the identification of diabetes retinopathy stages. J Med Syst (in press), 2008.
7.Gonzalez, R.C. and Wintz, P. Digital Image Processing. Second edition, AddisonWesley, Reading, MA, 1987.
8.Nayak, J., Bhat, P.S., Acharya, U.R., Lim, C.M., and Kagathi, M. Automated identification of different stages of diabetic retinopathy using digital fundus images. J Med Syst 32:107–115, 2008.
9.Wong, L.Y., Acharya, U.R., Venkatesh, Y.V., Chee, C., Lim, C.M., and Ng, E.Y.K. Identification of different stages of diabetic retinopathy using retinal optical images. Inf Sci 178:106–121, 2008.
10.Losch, B. Application of fuzzy sets to the diagnosis of glaucoma. Proceedings of the 18th Annual International Conference of the IEEE Engineering in Medicine and Biology 4:1550–1552, 1996.
11.Ulieru, M., Cuzzani, O., Rubin, S.H., and Ceruti, M.G. Application of soft computing methods to the diagnosis and prediction of glaucoma. Proceedings on IEEE International Conference on Systems, Man and Cybernetics 5:3641–3645, 2000.
225
Rajendra Acharya U. et al.
12.Parfitt, C.M., Mikelberg, F.S., and Swindale, N.V. The detection of glaucoma using an artificial neural network. Proceedings of 17th Annual Conference IEEE Engineering in Medicine and Biology 1:847–848, 1995.
13.Chan, K., Lee, T.-W., Sample, P.A., Goldbaum, M.H., Weinreb, R.N., and Sejnowski, T.J. Comparison of machine learning and traditional classifiers in glaucoma diagnosis.
IEEE Trans Biomed Eng 49(9):963–974, 2002.
14.Nayak, J., Acharya, U.R., Bhat, P.S., Shetty, A., and Lim, T.C. Automated diagnosis of glaucoma using digital fundus images. J Med Syst 33:337–346, 2009.
15.Walter, T., Klein, J.C., Massin, P., and Erginay, A. A contribution of image processing to the diagnosis of diabetic retinopathy-detection of exudates in color fundus images of the human retina. IEEE Trans Med Imaging 21:1236–1243, 2002.
16.Greaney, M.J., Hoffman, D.C., Garway-Heath, D.F. et al. Comparison of optic nerve imaging methods to distinguish normal eyes from those with glaucoma. Invest Ophthalmol Vis Sci 43:140–145, 2002.
17.Reynolds, D.A., Quatieri, T., and Dunn, R. Speaker verification using adapted Gaussian mixture models. Digital Signal Process 10:19–41, 2000.
18.Nelwamondo, F.V. and Marwala, T. Faults detection using Gaussian mixture models, mel-frequency cepstral coefficients and kurtosis. Systems, Man and Cybernetics SMC apos. IEEE International Conference 1:290–295, 2006.
19.Bizios, D., Heijl, A., and Bengtsson, B. Trained artificial neural network for glaucoma diagnosis using visual field data: a comparison with conventional algorithms. J Glaucoma 16:20–28, 2007.
20.Bowd, C., Chan, K., Zangwill, L.M., Goldbaum, M.H., Lee, T.W., Sejnowski, T.J., and Weinreb, R.N. Comparing neural networks and linear discriminant functions for glaucoma detection using confocal scanning laser ophthalmoscopy of the optic disc.
Invest Ophthalmol Vis Sci 43:3444–3454, 2002.
226
Chapter 7
Temperature Distribution Inside the Human Eye with Tumor Growth
Ooi, E.H. and Ng, E.Y.K.
7.1. Introduction to Temperature Distribution
In this study, we create a three-dimensional (3D) model of the human eye and investigate the thermal effects of an eye tumor on ocular temperature distribution. The steady state heat diffusion equation governs temperature inside a healthy ocular region, while the Pennes bioheat equation describes the heat flow inside the eye tumor. The eye model is numerically examined using the boundary element method, and the presence of an eye tumor is found to produce a warmer ocular temperature distribution. A slight thermal asymmetry is observed on the corneal surface. A parametric optimization based on the ANOVA and Taguchi methods is carried out to determine the importance and precedence of the various factors that may affect the ocular surface temperature and to determine the optimal setting of factors that maximize the signal from the eye tumor while isolating the other effects that may be classified as noise.
The temperature of the human body depends on its physiological condition. Around 400 b.c.e., the Greek physician Hippocrates posited that
diseases are likely to be detected in the parts of the human body where an excess of heat or cold is found. Modern medicine still relies on temperature
School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798.
227
Ooi, E.H. and Ng, E.Y.K.
monitoring for human health, as illustrated in the common practice of measuring body temperature for fever to determine if a patient is healthy or not. The application of thermal diagnostics extends beyond general practitioners. In breast cancer studies, medical researchers have made successful diagnoses based on the abnormal thermal patterns of the infected breast using IR thermography.1,2 The application of such thermal diagnostics systems has also been demonstrated in other diseases, such as vascular disorders,3,4 rheumatism,5−7 and the diseases that cause fever, such as SARS6 and bird flu.
The use of ocular temperature to monitor ocular physiology is well documented.8,9 Ocular abnormalities such as dry eye syndrome and acute inflammation have been found to produce ocular surface temperatures that are different from the temperature of a normal human eye.8−11 In more severe ocular diseases, such as cataracts, glaucoma, diabetic retinopathy, and eye tumors, no conclusive evidence has suggested that ocular surface temperature may be used as a means for detection. Nonetheless, extensive research and preliminary experimental investigations currently being carried out on patients with glaucoma have revealed promising results.12,13
In this study, a mathematical investigation is carried out to examine the thermal effects of eye diseases on the temperature distribution inside the human eye. We seek to determine the changes in ocular temperature distribution. In particular, we seek to measure changes in the corneal surface temperature and to determine the likelihood that these temperature changes will be detected using the current method of ocular thermometry.
Various ocular diseases may infect the human eye. However, in the present study, we shall limit our focus to eye tumors. One of the main reasons eye tumors are selected for the investigations carried out in this study is the availability of a well-established method for simulating the thermal behavior of tumors in biological tissues.14,15 According to Gautherie et al.16 tumor tissues have a higher rate of metabolic activity that leads to the production of a greater amount of metabolic heat. Additionally, the blood perfusion rate inside tumor tissues has been found to be, in general, greater than the blood perfusion rate of surrounding healthy tissues.17 These characteristics of tumor tissue enable the thermal effects of tumors to be simulated without much difficulty.
228
Temperature Distribution Inside the Human Eye with Tumor Growth
Despite the growing number of mathematical models that investigate the temperature distribution inside the human eye during various ocular circumstances, to the best of our knowledge, no mathematical studies have looked into the effects of an eye tumor on human eye temperature. On the other hand, investigations on the thermal effects of tumors that grow on other parts of the human body, such as the breast and skin, have been carried out by various researchers. Ng and Sudharsan14,15 developed a finite element model of the human breast to simulate the changes in temperature caused by the growth of a breast tumor. Similar studies have been reported by Hu et al.18 using the finite volume method. Models of the human skin with a tumor growing inside have been developed by Deng and Liu19,20 using various numerical approaches, such as the dual reciprocity method and the Monte-Carlo method.
In the models of the human breast and skin developed by Ng and Sudharsan,14,15 Hu et al.18 and Deng and Liu,19,20 each tumor was modeled as a homogeneous region with distinguishable thermal properties, characterized by a larger metabolic heat generation and blood perfusion rate when compared to the healthy surrounding tissue. Using these modeling approaches, the temperatures of the infected tissues were found to be 0.5–2◦C greater than the temperatures of healthy tissues.
In the present study, the same approach used for simulating the thermal behavior of tumors in human breast and skin cells is adopted to investigate the thermal effects of eye tumors on ocular temperature distribution. Simulations are carried out using a 3D model of the human eye. The boundary element method is used to obtain a numerical solution. The presence of terms associated with the metabolic heat generation and blood perfusion rate of the tumor in the governing bioheat equation gives rise to domain integrals in the integral representation derived. These integrals are managed using the dual reciprocity method.21
A parametric optimization based on the analysis of variance (ANOVA)22 and the Taguchi method23 is carried out to identify the precedence and importance of the various factors that may affect the ocular surface temperature and to determine the optimal setting of factors that maximize the signal from the eye tumor while isolating the other effects that may be classified as noise.
229