A Novel Approach on Automatic Detection of Optic Disc and Optic Cup Segmentation

(1)

ITSI Transactions on Electrical and Electronics Engineering (ITSI-TEEE)

_______________________________________________________________________________________________

ISSN (PRINT) : 2320 – 8945, Volume -4, Issue -2, 2016 19

A Novel Approach on Automatic Detection of Optic Disc and Optic Cup Segmentation

1Kavyashree M, ²P V Rao

1,2Dept of Electronics and Communication Engg., Rajarajeswari College of Engg., Bangalore-74,India

Abstract—This paper proposed a novel approach on Automatic detection of optic disc and optic cup segmentation glaucoma assessment by combined analysis of fundus eye image and patient data. Fundus image feature extraction and ocular parameter evaluation are carried out for image level analysis. The techniques used for feature extraction include color model analysis, morphological processing, filtering and thresholding.

Ocular parameters considered are Cup to Disc Ratio (CDR), Rim to Disc Ratio (RDR), cup to disc area ratio and Inferior Superior Nasal Temporal(ISNT) ratio of blood vessels in disc region. The CDR, RDR and cup to disc area ratio based on optic disc, cup and rim are calculated using image measuring techniques.

Keywords- CDR,RDR,ISNT, SVM Classify, Glaucomacup to disc ratio, morphological operation.

I. INTRODUCTION

This paper described a novel approach towards automatic glaucoma assessment. The efforts taken to develop an automatic glaucoma assessment technique have an integration of fundus image analysis and patient data analysis. Finally a combined glaucoma risk analysis is performed and a risk class is labelled for each set of input.

In fundus image analysis fundus image is preprocessed and features are extracted. The feature extraction involves optic disc segmentation, optic cup segmentation, optic rim segmentation and blood vessel extraction. An array centroid method is proposed to segment the extracted features into ISNT quadrants. Then ocular parameters such as CDR, cup to disc area ratio, Inferior RDR, Superior RDR and ISNT Ratio are calculated. [1]

Optic disc segmentation is one of the most important pre- processing steps in building an accurate automated glaucoma detection system. Glaucoma is one of the major causes for blindness, which cannot be reversed back. Once the nerves are damaged, it cannot be reconstructed back with our current level of technology.

However, early screening will allow the ophthalmologist to identify the disease during the early stages. A prevention step can be taken to limit the damage and save the remainder healthy nerves.

Thus, an automated system will be very beneficial to the public so that they can do early screening easily through the online system. Cup-to-disc ratio (CDR) is a method that utilized fundus camera modality to capture the back wall of the retina so that the ratiobetween cup disc and optics disc can be measured. The size of the optic disc will remain relatively stable for both healthy and glaucoma patient but the cup disc size will increase

proportionally to the severity of the glaucoma disease.

Thus, optic disc size needs to be determined before the ratio can be calculated.

Furthermore, Glaucoma is a chronic eye disease that leads to vision loss. As it cannot be cured, detecting the disease in time is important. Current tests using intraocular pressure (IOP) are not sensitive enough for population based glaucoma screening. Optic nerve head assessment in retinal fundus images is both more promising and superior. This paper proposes optic disc and optic cup segmentation using super pixel classification for glaucoma screening. In optic disc segmentation, histograms, and center surround statistics are used to classify each super pixel as disc or non-disc.

A self-assessment reliability score is computed to evaluate the quality of the automated optic disc segmentation. For optic cup segmentation, in addition to the histograms and center surround statistics, the location information is also included into the feature space to boost the performance. [3]

Fig 1.Medical Diagram of an Eye

II. LITERATURE SURVEY

Gayathri.R et.al [1] this paper proposed a method Glaucoma is an eye disease which damages the optic nerve of the eye and becomes severe over time. It is caused due to buildup of pressure inside the eye.

Glaucoma tends to be inherited and may not show up until later in life. The detection of glaucomatous progression is one of the most important and most challenging aspects of primary open angle glaucoma (OAG) management.

GopalDatt Joshi et.al [2] described a depth discontinuity (in the retinal surface)-based approach to estimate the approach shifts focus from the cup region used by existing approaches to cup boundary. The given sets of images, acquired sequentially, are related via a relative motion model and the depth discontinuity at the cup boundary is determined from cues such as motion boundary and partial occlusion.

(2)

_______________________________________________________________________________________________

MohdAsyrafZulkifley et.al [3] proposed Optic disc segmentation is a crucial step in automated glaucoma detection system through Cup-to-Disc ratio measurement.

Recent approaches focus on deterministic algorithm of RGB or grey model only. In this paper, we proposed a statistically integrated approach by combining various colour models. The driving motivation is the ability of each colour model t work accurately in certain environments or cases.

A.Murthi et.al [4] this method Proposed a method The optic cup-to-disc ratio (CDR) in retinal fundus images is one of the principle physiological characteristics in the diagnosis of glaucoma. The least square fitting algorithm aims to improve the accuracy of the boundary estimation.

The technique used here is a core component of ARGALI (Automatic cup-to-disc Ratio measurement system for Glaucoma detection and Analysis), a system for automated glaucoma risk assessment.

III. PROPOSED SYSTEM

The block diagram of the proposed architecture is shown in fig1. First we collect the retinal images from the medical data base. Then in the pre-processing system retinal images are converted into red component, green component. Red component is used to extract optic disc and cup areas for detecting Glaucoma. Then extracted optic disc is segmented in disc segmentation, by using the segmented disc enhance the vein region of retina.

And divide the enhanced vein region using graph nets and finally display the segmented result.[4,5]

A. Pre-Processing:

Pre-processing is an initial stage where the input MRI image will be taken and resized to 256X256 and convert it to gray image for further processing.

Input Retinal Image Pre-Processing Disc

Segmentation Cup Segmentation

Disc to Cup Ratio

Calculation Glaucoma Analysis Output Result

Figure 2: Architecture of Proposed Architecture A. Disc Segmentation

In optic disc segmentation, adaptive histogram equalization technique is used to improve the contrast of a colour image and Gabor filters are used for texture feature extraction. Combination of these two parameters is very useful for classification of each super pixel as disc or non-discsuper pixel.

B. Cup Segmentation

Optic cup segmentation, thresholding and binarization can be used. In binarization, depending upon the mean value of the entire pixel, each pixel in an image is assigned the value as „1‟ or „0‟ . If it is greater than the mean value then it is '1' else it is '0'.From a gray scale image, thresholding process is used to create binary images. In this process, if pixel value is greater than

some threshold value then that particular pixel in an image is marked as "object" pixels or else it will be considered as "background" pixels [8]. Typically, an object pixel is assigned a value of “1” whereas a background pixel is assigned a value of “0.”[6, 7]

Normal and abnormal retinal images are collected from different hospital of different patients for glaucoma analysis, early detection of glaucoma helpful for slow down the progression of disease.Input image is stored in the JPEG image format. Input RGB image is resize to 256x256 and double the image for higher precision then converted to grayscale image by eliminating hue and saturation and retaining luminance. Green channel is used for further processing and which contain more information compared to the other blue and red channel.

Green channel image is used for optic cup and red channel is used for the optic disc segmentation.

Green channel image is applied to the median filter which removes the noise and smoothens the edge.

Filtered output image is applied to the morphological operations which contain closing, opening, dilation and erosion operation. Close operation is a dilation followed by erosion which fill the gap in that area and by smoothing their edges. Image open operation is erosion followed by dilation. Output image of morphological operation is applied to watershed segmentation to extract the boundaries.

Start

Input Image

RGB to NTSC

Segmentation Green Channel

Morphological Operation

Cup Area Disc Area

Cup to Disc Ratio

CDR>03 Normal

Abnormal

Stop No

yes

Fig 3.Flowchart 1 of Morphological operation Images are stored in JPEG format .The original (RGB and NTSC) image is transformed into appropriate color space for further processes. Optic cup and disc is

(3)

_______________________________________________________________________________________________

segmented by using morphological operations, Hough transform and water shed segmentation. [8, 9]

Start

RGB to gray Scale Image

Adaptive Histogram Equalization Technique

Hough Transform

Cup Area Disc area

CDR

CDR>0.3

Normal Abnormal

yes

No

Fig 4.Flowchart 2 of Hough transform

IV. RESULTS

Images are stored in JPEG format .The original (RGB and NTSC) image is transformed into appropriate color space for further processes. Optic cup and disc is segmented by using morphological operations, Hough transform and water shed segmentation.

(a) (b) (c) Fig.2.(a)Original image (b)Optic disk (c)Optic cup

Table 1: CDR measurement for normal data base SL.

NO Cup Area

Disc Area

Cup to Disc Ratio

Data Base

Result

1 295.625 884.3750 0.3343 N N 2 366.75 1133.9 0.3234 N N 3 264.875 759.6250 0.3487 N N

4 260.870 794 0.3286 N N

5 577.25 1230.8 0.4690 N Abn 6 291.25 924.875 0.3149 N N

7 287.625 944 0.3047 N N

8 284.875 1044.8 0.2727 N N 9 273.25 1264.6 0.2761 N N 10 189.625 939.625 0.2018 N N 11 321.50 1192.5 0.2696 N N 12 206.5 982.25 0.2102 N N 13 187.625 1106.3 0.1696 N N 14 199.750 1055 0.1893 N N

15 205.875 707 0.2912 N N

Table 2:CDR measurement for abnormal data base SL.

NO

Cup area Disc area

CDR Data base

Res ult 1 565.25 1134.4 0.4983 Abn Abn 2 557.5 749.75 0.7439 Abn Abn 3 558.1250 905.375 0.6165 Abn Abn 4 652.1250 808.75 0.8063 Abn Abn 5 214.75 835.875 0.2569 Abn N

6 363.375 903 0.4204 Abn Abn

7 326.75 891.875 0.3836 Abn Abn 8 4746.1250 827.875 0.5751 Abn Abn 9 565.25 1134.4 0.4983 Abn Abn 10 557.7500 749.75 0.7439 Abn Abn 11 395.625 890.5 0.4443 Abn Abn 12 284.875 695.625 0.4095 Abn Abn

13 318 800 0.3975 Abn Abn

14 413.125 767.125 0.5387 Abn Abn 15 462.625 956.625 0.4460 Abn Abn Table 3:Glaucoma analysis

SL.N O

VC D

VD D

CDR=VCD/V DD

Glaucom a analysis

1 59 148 0.3986 Yes

2 65 146 0.4452 Yes

3 40 91 0.4396 Yes

4 39 78 0.500 Yes

5 30 80 0.260 No

6 31 80 0.3875 Yes

7 29 75 0.3867 Yes

8 48 135 0.3555 Yes

9 49 107 0.4579 Yes

(4)

_______________________________________________________________________________________________

V CONCLUSION

The cup to disc ratio is an important indicator of the risk of the presence of glaucoma in diabetes patients. In this paper the segmentation of optic disc, optic cup and smooth their boundaries by morphological operations will be used. The morphological operations are efficient to detect the cup to disc ratio in glaucoma patients and normal patients and then check the level of disease. If the cup to disc ratio is more 0.3 then those patients are glaucoma patients and if the disc ratio is less than those are normal patient .This operations has been tested on a different images. Flow chart 1 of morphological operation achieved 93.33% accuracy. Flow chart 2 of Hough transform achieved 85% accuracy.

REFERENCE

[1] Gayathri. R, Dr. P.V. Rao, Anma.S, “Automated Glaucoma Detection System based on Wavelet Energy features and ANN”.

[2] Gopal Datt Joshi, “Depth Discontinuity-Based Cup Segmentation From Multiview Color Retinal Images” Ieee Transactions On Biomedical Engineering, Vol. 59, No. 2012.

[3] Jun Cheng, Fengshou Yin, Damon Wing Kee Wong, “Sparse Dissimilarity-constrained Coding for Glaucoma Screening”, Ieee Transactions On Biomedical Engineering, 2015.

[4] MohdAsyrafZulkifley, AiniHussain, Mohd.

Marzuki Mustafa, Aouache Mustapha, “On

Analyzing Optic Disc Extraction through Weighted Colour Models”.

[5] A.Murthi& 2M.Madheswaran, “Enhancement Of Optic Cup To Disc Ratio Detection In Glaucoma Diagnosis”, 2012 International Conference on Computer Communication and Informatics (ICCCI) 2012.

[6] P.V.Raoa*, Gayathri.Rb, Sunitha.Rc, “A Novel Approach for Design and Analysis of Diabetic Retinopathy Glaucoma Detection using Cup to Disk Ration and ANN, 2014.

[7] D. Jeyashree, Ms. G. Sharmila , Dr. K. Ramasamy,

“ Combined Approach on Analysis of Retinal Blood Vessel

[8] Segmentation for Diabetic Retinopathy and Glaucoma Diagnosis”, Volume 5, Issue 5,2014.

Mohd Asyraf Zulkifley, Aini Hussain, Mohd.

Marzuki Mustafa, “On Analyzing Optic Disc Extraction through Weighted Colour Models”.

[9] Jun Cheng*, Jiang Liu, Yanwu Xu, Fengshou Yin, Damon Wing Kee Wong, Ngan-Meng Tan, Dacheng Tao, Ching-Yu Cheng, Tin Aung, and Tien Yin Wong.

[10] GopalDatt Joshi, Jayanthi Sivaswamy, and S. R.

Krishnadas, “Optic Disk and Cup Segmentation from Monocular Colour Retinal Images for Glaucoma Assessment”.

