Michael Dessauer and Sumeet Dua

matrix of second-order derivatives of L(x, y; t). For each scale t and location (x, y), we find:

We then use these values to calculate the determinant of the Hessian using

which can then be searched using a nonmaximal suppression algorithm to find scale-space localized, rotationally invariant features.31 These locations in scale-space represents locations that have “corners,” or large gradients in both x and y directions.

Retinal mosaicing, using the above methods to find location and scale, extracts a feature vector of binned gradients from the 2D Haar wavelet responses that are then used to match images for registration (Fig. 2.28).32 Scale space provides a unique multi-level view of images, which exploit scale-specific features.

2.5. Summary

We have presented many of the most popular methods of retinal image preprocessing, segmentation/localization, and feature extraction for automated clinical decision support. Methods that are more complex have been introduced and can provide further guidance of how advanced methods, such as steerable filters,33 active contours,34 and neural networks,35 are constructed. A retinal image classification algorithm can be assessed and observed as a series of steps consisting of many of the methods described in this chapter. It is important for the researcher/scientist to attempt multiple methods at each stage in the algorithm to best assess how the method affects the overall performance of the algorithm. Although the ultimate goal of a fully automated decision support system for retinal pathology detection has not been fully realized, there continue to be great strides toward reaching this challenge through computational methods.

84

Computational Methods for Feature Detection in Optical Images

References

1.Winder, R.J., Morrow, P.J., McRitchie, I.N., Bailie, J.R., and Hart, P.M. Algorithms for digital image processing in diabetic retinopathy. Comput Med Imaging Graph 33:608– 622, 2009.

2.Frame, A.J., Undrill, P.E., Cree, M.J., Olson, J.A., McHardy, K.C., Sharp, P.F., and Forrester, J.V. A comparison of computer based classification methods applied to the detection of microaneurysms in ophthalmic fluorescein angiograms. Comput Biol Med 28:225–238, 1998.

3.Joshi, G.D. and Sivaswamy, J. Colour retinal image enhancement based on domain knowledge. ICVGIP’08: Proceedings of the 2008 Sixth Indian Conference on Computer Vision, Graphics & Image Processing. Washington, DC, USA: IEEE Computer Society,

pp.591–598, 2008.

4.Hoover, A. and Goldbaum, M. Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels. IEEE Trans Med Imaging 22:951–958, 2003.

5.Wang, H., Hsu, W., Goh, K.G., and Lee, M.L. An effective approach to detect lesions in color retinal images. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit,

pp.1–6, 2000.

6.Spencer, T., Olson, J.A., McHardy, K.C., Sharp, P.F., and Forrester, J.V. An imageprocessing strategy for the segmentation and quantification of microaneurysms in fluorescein angiograms of the ocular fundus. Comp Biomed Res 29:284–302, 1996.

7.Foracchia, M., Grisan, E., and Ruggeri, A. Luminosity and contrast normalization in retinal images. Med Image Anal 9:179–190, 2005.

8.Gonzalez, R.C. and Woods, R.E. Digital Image Processing, Prentice Hall, 2002.

9.Sinthanayothin, C., Boyce, J.F., Cook, H.L., and Williamson, T.H. Automated localization of the optic disc, fovea, and retinal blood vessels from digital color fundus images. Br J Ophthalmol 83:902–910, 1999.

10.Shapiro, L.G. and Stockman, G.C. Computer Vision, pp. 137–150, Prentice Hall, 2001.

11.Newsom, R.S.B., Sinthanayothin, C., and Boyce, J. Contrast enhancement for oral fluorescein angiograms. BJOP 1:29–32, 1998.

12.Goatman, K.A., Whitwam, A.D., Manivannan, A., Olson, J.A., and Sharp, P.F. Color normalization of retinal images. Proceedings Medical Image Understanding and Analysis, pp. 49–52, 2003.

13.Rapantzikos, K. Detection and segmentation of drusen deposits on human retina: potential in the diagnosis of age-related macular degeneration. Med Image Anal 7:95–108, 2003.

14.Yulong, M. and Dingru, X. Recognizing the glaucoma from ocular fundus image by image analysts. Engineering in Medicine and Biology Society, Proceedings of the Twelfth Annual International Conference of the IEEE, pp. 178–179, 1990.

15.Gregson, P., Shen, Z., Scott, R., Kozousek, V. Automated grading of venous beading.

Comput Biomed Res 28:291–304, 1995.

16.Spencer, T., Olson, J.A., McHardy, K.C., Sharp, P.F., and Forrester, J.V. An imageprocessing strategy for the segmentation and quantification of microaneurysms in fluorescein angiograms of the ocular fundus. Comp Biomed Res 29:284–302, 1996.

85

Michael Dessauer and Sumeet Dua

17.Pinz, A., Bernogger, S., Datlinger, P., and Kruger, A. Mapping the human retina. IEEE Trans Med Imag 17:606–619, 2002.

18.Yulong, M. and Dingru, X. Recognizing the glaucoma from ocular fundus image by image analysts. Engineering in Medicine and Biology Society Proceedings of the Twelfth Annual International Conference of the IEEE pp. 178–179, 1990.

19.Sinthanayothin, C., Boyce, J.F., Williamson, T.H., Cook, H.L., Mensah, E., Lal, S., and Usher, D. Automated detection of diabetic retinopathy on digital fundus images. Diabetic Med 19:105–112, 2002.

20.Meyer, F. Topographic distance and watershed lines. Signal Processing 38:113–125, 1994.

21.Walter, T. and Klein, J.-C. Segmentation of color fundus images of the human retina: detection of the optic disc and the vascular tree using morphological techniques. Medical Data Analysis. Crespo, J., Maojo,V., and Martin, F. (Ed.), Lecture Notes in Computer Science 2199:282–287, 2001.

22.Chaudhuri, S., Chatterjee, S., Katz, N., Nelson, M., and Goldbaum, M. Detection of blood vessels in retinal images using two-dimensional matched filters. IEEE Trans Med Imag 8:263–269, 1989.

23.Hipwell, J.H., Strachan, F., Olson, J.A., McHardy, K.C., Sharp, P.F., and Forrester, J.V. Automated detection of microaneurysms in digital red-free photographs: a diabetic retinopathy screening tool. Diabetic Med 17:588–594, 2000.

24.Haralick, R.M., Shanmugan, K., and Dinstein, I. Textural features for image classification. IEEE Trans Syst Man Cybern SMC-3:610–621, 1973.

25.Hu, M.K. Visual pattern recognition by moment invariants. IRE Trans Info Theory IT-8:179–187, 1962.

26.Cooley, J.W. and Tukey, J.W. An algorithm for the machine computation of the complex Fourier series. Math Comput 19:297–301, 1965.

27.Tan, W., Wang, Y., and Lee, S. Retinal blood vessel detection using frequency analysis and local-mean-interpolation filters. Proceedings of SPIE Med Imaging: Image Processing 4322:1373–1384, 2001.

28.Pearson, K. On lines and planes of closest fit to systems of points in space. Philosophical Magazine 2(6):559–572, 1901.

29.Mallat, S. A Wavelet Tour of Signal Processing, Academic Press, New York, 1999.

30.Lalonde, M., Beaulieu, M., and Gagnon, L. Fast and robust optic disc detection using pyramidal decomposition and Hausdorff-based template matching. IEEE Trans Med Imag 20:1193–1200, 2001.

31.Bay, H., Ess, A., Tuytelaars, T., and Vangool, L. Speeded-up robust features (SURF).

Computer Vision and Image Understanding 110:346–359, 2008.

32.Cattin, P., Bay, H., Van Gool, L., and Székely, G. Retina mosaicing using local features. Medical Image Computing and Computer-Assisted Intervention — MICCAI 2006. Larsen, R., Nielsen, M., and Sporring, J. (Ed.), Lecture Notes in Computer Science 4191:185–192, 2006.

33.Kochner, B., Schuhmann, D., Michaelis, M., Mann, G., and Englmeier, K.-H. Course tracking and contour extraction of retinal vessels from color fundus photographs: most efficient use of steerable filters for model-based image analysis. Society of

86

Computational Methods for Feature Detection in Optical Images

Photo-Optical Instrumentation Engineers (SPIE) Conference Series. Hanson, K.M. (Ed.), pp. 755–761, 1998.

34.Saradhi, G.V., Balasubramanian, S., and Chandrasekaran, V. Performance enhancement of optic disc boundary detection using active contours via improved homogenization of optic disc region. International Conference on Information and Automation pp. 264– 269, 2006.

35.Gardner, G., Keating, D., Williamson, T.H., and Ell, A.T. Automatic detection of diabetic retinopathy using an artificial neural network: a screening tool. Br J Ophthalmol 80:940–944, 1996.

87

Chapter 3

Computational Decision

Support Systems and Diagnostic

Tools in Ophthalmology:

A Schematic Survey

Sumeet Dua and Mohit Jain

Computer decision support systems are computer applications that can help clinicians make diagnostic decisions for patient treatment. Computer decision support systems provide easy access to patient file repositories. By accessing the data from the computational system, clinicians can decide whether to opt for surgery, therapy, or another form of treatment. Moreover, the system can alert clinicians to new patterns in patient data. Because of these benefits, a computer decision support system can save a clinician’s time and can be more effective than the clinician can in making treatment decisions. Designing a computational system is not easy because the software and hardware infrastructure requirements are complicated and may negatively affect the computer’s user friendliness. Additionally, training clinicians for this computational system can be costly. Despite these difficulties, the benefit and convenience that this system provides surpasses those offered by the old methods. This automatic and interactive system can help clinicians make better clinical decisions, which will help them to treat patients more effectively by helping clinicians avoid risks.

Department of Computer Science, College of Engineering and Science, Louisiana Tech University, Ruston, LA, U.S.A.

89