Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched...
-
Upload
chrystal-joseph -
Category
Documents
-
view
222 -
download
0
Transcript of Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched...
![Page 1: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/1.jpg)
Fast Localization and Segmentation of Optic Disk inRetinal Images Using Directional Matched Filtering
and Level Sets
Project Guide/Co-Guide:
P.Rekha Sharon, M.E(AE)Register No:312211401014
Mr.C.VinothKumar
![Page 2: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/2.jpg)
Objective
Diabetic Macular Edema (DME) is a threatening
complication of diabetic retinopathy.
Severity of DME can be assessed by detecting exudates (a
type of bright lesions) in color fundus images. This can be
done by separating the Optical Disk (OD) from the retinal
image.
To identify the location of the optical disc to prevent the
confusion made during finding the infected region.
![Page 3: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/3.jpg)
Motivation
OD localization techniques suffer from impractically-high
computation times.
In this work, they present a fast technique that requires less
time to localize the OD.
The technique is based upon obtaining two projections of
certain image features that encode the x- and y- coordinates of
the OD.
![Page 4: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/4.jpg)
Proposed Algorithm
![Page 5: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/5.jpg)
OPTICAL DISC SIZE ESTIMATION
• An important parameter that needs to be determined in our OD detection and segmentation algorithm is the size of the OD.
fimg = AFOV
NFOV
AFOV-Area of FOV
NFOV-No of Pixels in FOV
![Page 6: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/6.jpg)
OD LOCALIZATION
• The steps are ,
Background normalization, Template matching, Directional matched filtering,
![Page 7: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/7.jpg)
Background Normalization• To reduce the false detection of OD candidates due to non-uniform
illumination.
• Expand the original image border by half of the filter window size.
• Each pixel’s intensity for the out-of-FOV dark region is replaced by averaging gray levels of pixels in the FOV.
• The purpose of the expansion is to remove the artifacts at the image border in the processed image due to the large size of the filter and the black background.
![Page 8: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/8.jpg)
![Page 9: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/9.jpg)
Template Matching
• The purpose of template matching is only to provide the OD candidate locations.
• To locate the OD candidates a binary template where the disk, given in white, is assigned a value 1 and the black background is assigned a value 0.
• The radius of the white circle in the template is the estimated OD radius ‘r’. The
![Page 10: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/10.jpg)
Directional matched filtering
• To remove false positives and locate OD centre by finding main vessel arcades origin.
![Page 11: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/11.jpg)
![Page 12: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/12.jpg)
Work Done in Phase-I
• Retinal Blood Vessels have been extracted from the Fundus Image by the following steps,
Image Preprocessing Image fFiltering Image Enhancements
Image SegmentationTexturingThresholdingMorphological Operations
Thinning
Dilation
![Page 13: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/13.jpg)
WORK DONE SO FAR IN PHASE-II
Completed Modules for
Optical Disc size estimation, OD localization,
Background normalization, Template matching, Directional matched filtering,
![Page 14: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/14.jpg)
WORK TO DO IN PHASE-II
Modules for OD Segmentation OD segmentation.
Image Preprocessing,Blood vessel removal,Bright region removal,Fast, Hybrid Level Set Model,Least-Square Ellipse Fitting.Enhancement or Detection of Exudates.
![Page 15: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/15.jpg)
References
[1] D. Usher, M. Dumskyj, M. Himaga, T. H. Williamson, S. Nussey, and J. Boyce, “Automated detection of diabetic retinopathy in digital retinal images:Atool for diabetic retinopathy screening,” Diabetic Med., vol. 21, pp. 84–90, 2004.
[2] A. D. Fleming, K. A. Goatman, S. Philip, G. J. Prescott, P. F. Sharp, and J. A. Olson, “Automated grading for diabetic retinopathy: A large-scale audit using arbitration by clinical experts,” Br. J. Ophthalmol., vol. 94, pp. 1606–1610, Dec. 2010.
[3] C. Agurto, V. Murray, E. Barriga, S. Murillo, M. Pattichis, H. Davis, S. Russell, M. Abramoff, and P. Soliz, “Multiscale AM-FM methods for diabetic retinopathy lesion detection,” IEEE Trans. Med. Imag., vol. 29, no. 2, pp. 502–512, Feb. 2010.
[4] K. A. Goatman, A. D. Fleming, S. Philip, G. J. Williams, J. A. Olson, and P. F. Sharp, “Detection of new vessels on the optic disc using retinal photographs,” IEEE Trans. Med. Imag., vol. 30, no. 4, pp. 972–979, Apr. 2011.
[5] G. D. Joshi, J. Sivaswamy, and S. R. Krishnadas, “Optic disk and cup segmentation frommonocular color retinal images for glaucoma assessment,” IEEE Trans. Med. Imag., vol. 30, no. 6, pp. 1192–1205, Jun. 2011.
[6] R. J. Winder, P. J. Morrow, I. N. McRitchie, J. R. Bailie, and P. M. Hart, “Algorithms for digital image processing in diabetic retinopathy,” Comput. Med. Imag. Graph., vol. 33, pp. 608–622, 2009.
![Page 16: Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets Project Guide/Co-Guide: P.Rekha Sharon,](https://reader037.fdocuments.net/reader037/viewer/2022110213/56649f355503460f94c5296c/html5/thumbnails/16.jpg)
Thank You!!!