Automated Detection of Optic Disc in Retinal FundusImages Using PCA

IOSR Journal of Pharmacy and Biological Sciences (IOSR-JPBS)
e-ISSN: 2278-3008, p-ISSN:2319-7676. Volume 10, Issue 6 Ver. III (Nov - Dec. 2015), PP 47-53
www.iosrjournals.org
DOI: 10.9790/3008-10634753 www.iosrjournals.org 47 | Page
Automated Detection of Optic Disc in Retinal FundusImages
Using PCA
Sangramsing Kayte 1
, Raju Maher 2
, Dr.Charansing Kayte3
1,2
Department of Computer Science & IT, Dr. Babasaheb Ambedkar Marathwada University, Aurangabad,
Maharashtra,
3
Assistant Professor of Digital and Cyber Forensic, Maharashtra.
Abstract: An automatic system is presented to find the locationof optic disc andit is a major landmark for the
detection of other anatomic features and the macula. These localizationare found by principal component
analysis to the image,that contains each structure.The detection of optic disc in colour retinal photograph is a
significant task in an automated retinal image analysis system. It is a challenging task to detect optic disc in the
normal as well as abnormal, that is, images affected due to disease. This paper presents an automated system to
detect an optic disc and its center in normal and abnormal retinal images. The proposed algorithm gives
excellent results of optic disc detection. The proposed technique is developed standard databases provided of
online Diaretdb0 130 images and Diaretdb1 89 images the images size of 1500 × 1152. The results achieved by
after tested images different algorithms can be compared when algorithms are applied on same standard
databases. This paper result comparison with another paper result and in which shows that the proposed
algorithm is well efficient.
Keywords: Retinal image, normal image; abnormal image; optic disc.
I. Introduction
Detection of normal and abnormal features in retinal images is important and helpful for automatic
detection of ophthalmologist. The normal features are the anatomical structures of fundus images that include
optic disk, macula, and fovea. It allows recording the diagnostic data and enabling the ophthalmology discussion
afterwards. For a particularly long time, automatic diagnosis of retinal diseases from digital fundus images has
been an active research topic in the medical image processing.Optic disk is a bright yellowish disk where the
blood vessels and the optic nerves occur. The dimensions of the optic disk are often studied for signs of some
diseases. To identify the contour of the optic disk a method based on pca contours with new
variationalpreparation is proposed.Fig. 1 shows main anatomical structures the retinal fundus image. The retina
is an interior part of eye which acts as the film of eye. It converts light rays into electrical signals and sends them
to the brain through the optic nerve. Optic nerve is the cable connecting the eye to the brain[1]. Optic disc is the
bright region within the retinal image. It is the spot on the retina where the optic nerve and blood vessels enter
the eye. Macula is responsible for our central vision. The fovea is an indentation in the centre of the macula.
This small part of our retina is responsible for our highest visual acuity. The vascular network is a network of
vessels that supply nutrients and blood to the retina.
An important criterion for automation is the accurate localization of the main anatomical features in the
image. An accurate and efficient detection of optic disc in an automated retinal image analysis system.
Fig. 1.ColourRetinal Fundus Image

Automated Detection of Optic Disc in Retinal FundusImages Using PCA
The optic disc localization is important for many reasons. The automatic and efficient detection of the
location of the optic disc in colour retinal images is an important and fundamental step in the automated retinal
image analysis system [1], [2].To successfully find abnormal structures in a retinal image An example of this is
the optic disc, an anatomical structure with a bright appearance, which should be ignored when detecting bright
lesions. The attributes of optic disc are similar to hard exudates in terms of colour and brightness. Therefore it is
located and removed during hard exudates detection process, thereby elapse false positives.
Optic disc detection is the main step to automated screening systemsdevloping for diabetic retinopathy
and glaucoma. Optic disc boundary and localization of macula are the two features of retina necessary for the
detection of exudates and also knowing the severity of the diabetic maculopathy [3]. In case of diabetic
maculopathy lesions identification, masking the false positive optic disc region leads to improvement in the
performance of lesion detection. The distribution of the abnormalities associated with some retinal diseases (e.g.
diabetic retinopathy) over the retina is not uniform; certain types of abnormalities more often occur in specific
areas of the retina [7]. The position of a lesion relative to the major anatomy could thus be useful as a feature for
later analysis. It is used as prerequisite for the segmentation of other normal and pathological features by many
researchers. The position of optic disc can be used as a reference length for measuring distances in retinal
images, especially for the location of macula. In case of blood vessel tracking algorithms, the location of optic
disc becomes the starting point for vessel tracking.
The optic disc, fovea, blood vessel bifurcations and crosses can be used as control points for registering
retinal images [8]. The registration of retinal images is an important step for superresolution and image change
detection. Unique feature points within image are used as control points for registration. Optic disc is a unique
anatomic structure within retinal image. These methods play major role in automatic clinical evaluation system.
When feature based registration algorithms are used, the accuracy of the features themselves must be considered
in addition to the accuracy of the registration algorithms [9]. Optic disc acts as landmark feature in registration
of multimopticdiscal or temporal images.
Location of the retinal optic disc has been attempted by several researchers recently. According to S.
Sekhar et al., the optic disc is usually the brightest component on the fundus, and therefore a cluster of high
intensity pixels will identify the optic disc location [10].
Sinthanayothin et al. [11] presented a method to detect the location of the optic disc by detecting the
area in the image which has the highest variation in brightness. As the optic disc often appears as a bright disc
covered in darker vessels, the variance in pixel brightness is the highest there. They also presented method for
the detection of the macular centre. They used a template matching approach in which the template was a
gaussian blob. The search area was constricted by the fact that the macular centre was assumed to be in the
darkest part of the image approximately 2.5 times the optic disc diameter from the optic disc [12]. In macula
localization the approximate distance between optic disc and macula is used as a priori knowledge for locating
the macula [13]. A method based on pyramidal decomposition and Handoff-distance based template matching
was proposed by Lalonde et al. [14]. The green plane of the original image was sub-sampled and the brightest
pixels in this sub-sampled image were selected as candidate regions. An edge detector was used on the candidate
regions in the original image. Next, multiple circular templates were fit to each of the regions using the
Hausdorff-distance as a distance measure. The centre of the fitted circular template was taken as the optic disc
centre.
Sopharak et al. [15] presented the idea of detecting the optic disc by entropy filtering. After pre-
processing, optic disc detection is performed by probability filtering. Binarization is done with otsu’s algorithm
[16] and the largest connected region with an approximately circular shape is marked as a candidate for the optic
disc. Hoover et al. [17] described a method based on a fuzzy voting mechanism to find the optic disc location. In
this method the vasculature was segmented and the vessel centrelines were obtained through thinning. After
removal of the vessel branches, each vessel segment was extended at both ends by a fuzzy element. The location
in the image where most elements overlap was considered to be the optic disc.
Ravishankar et al. [18] tried to track the optic disc by combining the convergence of the only thicker
blood vessels initiating from it and high disk density properties in a cost function. A cost function is defined to
obtain the optimal location of the optic disc that is a point which maximizes the cost function.
Niemeijer et al. [19] defines a set of features based on vessel map and image intensity, like number of
vessels, average width of vessels, standard deviation, orientation, maximum width, density, average image
intensity etc. The binary vessel map obtained [20] is thinned until only the centerlines of the vessels remain and
all the centerline pixels that have two or more neighbors are removed. Next, the orientation of the vessels is
measured by applying principal component analysis on each centerline pixel on both sides. Using the circular
template of radius 40 pixels having manually selected optic disc center within the radius, all features are
extracted for each sample location of the template including distance d to the true centre. To locate optic disc, a
sample grid is overlaid on top of the complete field of view and features vector are extracted and location of
optic disc is found containing pixels having lowest value of d.

Improved results on the same dataset were reported by Foracchia et al. [21]. They described a method
based on the global orientation of the vasculature. A simple geometrical mopticdiscel of the average vessel
orientation on the retina with respect to the optic disc location was fitted to the image.
In most of the papers researchers considered the optic disc as the brightest region within retinal image.
However, this criterion may not be applicable for retinal images those include other bright regions because of
diseases such as exudates due to diabetic retinopathy. Some considered the optic disc as the area with highest
variation in intensity of adjacent pixels. Both the criteria considered by many researchers are applicable for
normal, healthy retinal images. M.D. Abramoff and M. Niemeijer clearly mentioned in the paper [2] that the
approach in this paper has the potential to detect the location of the optic disc in retinal images with few or no
abnormalities.
This paper presents a novel algorithm for optic disc localization. The proposed algorithm groups the
steps based on different principles and propticdiscuces more accurate results. First we estimated threshold using
green channel histogram and average number of pixels occupied by optic disc. Applying this threshold, all bright
regions within image called clusters are detected. Then we applied two different criteria on these clusters, a: area
criterion and b: density criterion. The details about this are discussed in further sections. Once the candidate
cluster for optic disc is identified, the brightest oculus criterion is applied to locate the centreof optic disc.
The propose an automatic system to locate an optic disc not only in normal, healthy images but also in
images affected because of diseases such as diabetic retinopathy and images of poorer quality. There are more
chances of false optic disc detection in images affected due to diseases and images of poor quality than
desirable. The problem with retinal images is that the quality of the acquired images is usually not gooptic disc.
As the eye-specialist does not have complete control over the patient’s eye which forms a part of the imaging
optical system, retinal images often contain artifacts and/or are of poorer quality than desirable [24]. Despite
controlled conditions, many retinal images suffer from non-uniform illumination given by several factors: the
curved surfaces of the retina, pupil dilation (highly variable among patients) or presence of disease among others
[25]. However, our system avoids detecting false optic disc applying different criteria based on different
principles. We tested proposed system on 453 retinal images which include normal (healthy) as well as
abnormal (affected) retinal images. We are able to locate optic disc in 98.45% of all tested cases. Once the optic
disc is located accurately, its centre is also located accurately.
II. Material And Methods
Database used for optic disc localization is as shown in Table I.
TABLE I DATABASE USED FOR OPTIC DISC LOCALIZATION
Sr.
No.
Test Database Number of
Images
1 Diaretdb0 130
2 Diaretdb1 89
Thus, a set of 219 retinal images is studied for automated localization of optic disc and its centre.
A.Proposed Algorithm for Detection of Candidate Region for
optic disc
Step 1: Estimate Threshold.
Step 2: Apply Threshold and identify bright regions. Step 3: Select candidate regions which satisfy area
criterion.
Step 4: Select candidate region which satisfies density criterion.
Step 5: If no candidate region is selected, reduce threshold.
Step 6: If threshold is greater than zero, apply steps 2 through 5.
Step 7: Stop.
The major steps in the algorithm are discussed in detail here.
1)Threshold estimation
In healthy retinal images, optic disc is mostly the brightest region. However, in retinal images affected
due to diseases such as diabetic retinopathy, there may other bright regions in addition to optic disc. So first we
detected all bright regions within retinal images. In paper by Li and Chutatape [22], they used the highest 1%
gray levels in intensity image to obtain threshold value to detect candidate bright regions. However, there is risk
of not detecting opticdisc as candidate region if highest 1% gray levels are occupied by other bright regions
within image.
Siddalingaswamy et al. [1] used iterative threshold method to estimation threshold for optic disc
detection. This criterion is also not applicable to all types of retinal images.
Green channel image shows better contrast than red channel or blue channel image. It is observed that
optic disc appears most contrasted in the green channel compared to red and blue channels in RGB image.

Therefore, only the green channel image is used for the effective thresholding of the retinal image. So, we
estimated threshold considering green channel histogram. Optimal thresholdingmethod divides the pixels of the
image in two groups: group A and group B such that group A contains pixels at least equal to the number of
pixels occupied by the optic disc. Optic disc size varies from person to person. It is a vertical oval, with
average dimensions of 1.76mm horizontally by 1.92mm vertically [26]. Its width and height are 1/8 and 1/7.33
of retinal image diameter, respectively [27]. Thus, it is possible to determine the number of pixels occupied by
the optic disc as :
where D is the diameter of the retinal image in pixels.
To obtain an optimal threshold, the histogram derived from the source image is scanned from highest
intensity value to the lowest intensity value. The scanning stops at intensity level T when scanned pixels are
greater than the estimated optic disc pixels and there is a 10% rise in pixel count between two consecutive
intensity levels. Thus, the optimal threshold is calculated as follows :
Step1 : Initialize i = 255 and sum = 0
Step2: sum = sum + H[i]
Step3: i = i 1
Step4: if sum count or
i 1 i
1repeat steps 2 through 4
[ i
Step5: Threshold, T = i
where H[i] indicates the histogram of the source image and
i indicates the intensity level.
2) Area criterion
The optimal threshold when applied to the image results in one or more isolated connected regions (clusters).
Each of the cluster in the thresholded image is labeled and total number of pixels in each cluster are calculated.
The clusters having more than 125% or less than 10% of the optic disc area are discarded. This criterion
minimizes the possibility to miss the optic disc from the selected candidate clusters.
(a) (b)
Fig. 2.(a) Original image (b)After Filtering
3) Density criterion
The density criterion is applied to clusters which have already satisfied the area criterion. According to
the density criterion, if the ratio of number of pixels occupied by cluster to the number of pixels occupied by
rectangle surrounding the cluster is less than 40%, the cluster is discarded. From the remaining clusters the
cluster having highest density is considered to be the primary region of interest. Fig. 4
A.Localization of Centre of an optic disc
The cluster which occupies optic disc is located in algorithm discussed above. The centroid of this
cluster is determined using calculus method. A search area is defined around this centroid such that this centroid
is center of the search window. This search window is a square window with side equal to twice of optic disc
diameter (optic discd).
A circular window called an oculus of radius optic discd/2 is moved across the search area. This is
illustrated in Fig. 6. The maximum intensity oculus is identified using procedure given here.

Fig. 3. Center Circular window for optic disc detection
Each pixel within the square window of side equal to optic disc is tested for its distance (d) from the
centre of the window. As shown in Fig.4, if the distance, d is less than or equal to the radius r (radius of the
oculus), it is considered as inside pixel. The total intensity of the oculus is calculated by adding squares of
intensities of all inside pixels.
The centre of maximum intensity oculus is marked as a centre of optic disc.
Fig. 4. Center for Pixel inside circle test
There may be more than one window of same maximum total intensity. In this case, the central window
amongst the same intensity windows is the resultant window and the centre of the resultant window is
considered as a centre of an optic disc.
Fig. 5.a)andExamples of centre of optic disc Detection.
III. Results
TABLE II RESULTS OF PROPOSED OPTIC DISC LOCALIZATION METHOD
Test database Images optic
disc
Detected
%
Accuracy
Diaretdb0 130 127 98.04
Diaretdb0 89 87 97.47
Total 219 214 97.75
The outcome of optic disc localization is deemed true detection if obtained centre is within the optic disc
area. Table II shows the accuracy of true optic disc detected retinal images.
(
b)a)

A.Comparison of results obtained using different methods
Rashid Jalal Qureshi et al. [28] discussed different OPTIC DISC detection algorithms and optic disc
percentage detection rate in each case using standard databases. Table III summarizes the percentage accuracy
achieved applying each algorithm discussed in this paper and proposed algorithm.
We can easily conclude observing Table III that proposed method gives better accuracy in localizing
optic disc compared to other methods. Rashid Jalal Qureshi et al. [28] mentioned that performance of the
methods listed in columns 2 through 6 of Table III is generally gooptic disc, but each method has situations,
where it fails. These methods fail on a difficult data set i.e., the diseased retinas with variable appearance of
optic discs in terms of intensity, colour, contour definition etc. The criteria used in the proposed algorithm are
determined by considering abnormality of retinal images and hence provides better accuracy in localizing optic
disc as shown in the last column of Table III.
IV. Discussion And Conclusion
This proposed automated method has been presented which is localization an optic disc in retinal
images. The results show that the system is able to locate the optic disc accurately in 97.75% of all tested cases.
The percentage of successful detection of optic disc is increased using method presented in this paper.
Themethod of optic disc localization is tested on retinal images and qualitatively valuated by comparing the
automatically segmented optic disc with center can be detected accurately. Manual ones detected by an
experienced ophthalmologist or clinicians in detecting and diagnosing retinal diseases. Compared to the
approaches by other researchers, our algorithm for optic disc detection has the advantage that it is applicable to
all types of retinal images, healthy as well as abnormal, affected due to disease and/or acquisition process.
The work in this paper as the optic disc and its center are located accurately, macula and center of optic
disc, center of macula can be used as control points for registration of retinal images. The bifurcation points
within retinal images which are to be registered can be correlated by checking their distances from centerof optic
disc and center of macula. The accurate registration of retinal images can be further used for retinal image.
References
[1] Siddalingaswamy C , Gopalakrishnarabhu K , “Automatic localization and boundary detection of optic disc using implicit active
contours”, Internation Journal of Computer Applications, vol. 1, no. 7, pp. 1-5, 2010.
[2] Michael D. Abramoff, MeindertNiemeijer, “The automatic detection of the optic disc location in retinal images using optic disc
location regression”, Confroc IEEEEng Med BiolSoc , 1: pp 44 2- 4435, 2006.
[3] Jaspreet Kaur, Dr Sinha, “Automated localization of optic disc and macula from fundus images”, International Journal of
AdvancedResearch in Computer Science and Software Engineering, vol. 2, Issue 4, pp. 242-249, April 2012.
[4] Healey PR, Mitchell P, Smith W, Wang JJ, “Relationship between cupdisc ratio and optic disc diameter”, the Blue Mountains Eye
Study,AustN Z J Ophthalmol.; 25 Suppl 1:S99-101, May 1997.
[5] The American Academy of Ophthalmology, The eye MD Association,San Francisco, CA 94120-7424.
[6] GopalDatta Joshi, JayanthiSivaswamy, “Optic disc and cup boundary detection using regional information”, Arvind Eye Care
System, Madurai, India.
[7] J Tang, S Mohr, YD Du and TS Kern, “Non-uniform distribution of lesions and biochemical abnormalities within the retina of
diabetic humans”, Current Eye Res , vol.27, no.1, pp. 7-13, 2003.
[8] Mads Bo Christensen, Christian Bork Hardahl, Michael MunkJakobsen, Toke SonnenburgOttesen and Sille Petersen, “Automatic
algorithm for segmentation, registration and fusion of digital fundus retinal images from patients suffering from diabetic
retinopathy”, Aalborg University, SEMCON, 21, Dec. 2007.
[9] R.J. Radke, SrinivasAndra, Omar Al-Kofahi, BadrinathRoysam,“Image change detection algorithms: A systematic survey”, IEEE
Transactions on Image Processing, vol. 14, no. 3, March 2005.
[10] S. Sekhar, W. Al-Nuaimy and A K Nandi, “Automated localization of retinal optic disc using Hough transform”, pp 1577-1580,
IEEE 2008.
[11] C. Sinthanayothin, J Boyce, Cook and T Williamson, “Automated localization of the optic disc, fovea and retinal blood vessels
from digital color fundus images, Br J Ophthalmol, 83: pp. 902-910, Feb. 1999.
[12] MeindertNiemeijer, Michael D. Abramoff, Bram Van Ginneken, “Segmentation of the optic disc, macula and vascular arch in
fundus photographs”, IEEE Transactions on medical Imaging, vol.26, no.1, pp. 116-127, January 2007.
[13] SiddalingaswamyC ,Gopalakrishnarabhu K , “Automated detection of anatomical structures in retinal images”, 7th IEEE
International Conference on computational intelligence and multi-media applications, vol. 3, pp. 164-168, 2007.
[14] M Lalonde, M Beaulieu and Langis Gagnon, “Fast and robust optic disc detection using pyramidal decomposition and Hausdorff-
Based Template Matching”, vol. 20, No.11, pp. 1193-2001, Nov. 2001.
[15] Sopharak A., Thet New K., Aye Moe Y., N. Dailey M., Uyyanonvara B.,“Automatic exudate detection with anaivebayes classifier”,
International Conference on Embedded Systems and Intelligent Technology, Bangkok, Thiland, pp. 139-142,2008.
[16] N. Otsu, "A threshold selection method from gray-level histograms," IEEE Transactions on Systems. Man, and Cybernetics, vol. 9,
pp 62-66, January 1979.
[17] A oover and M Goldbaum, “Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels”, IEEE
Trans Med. Image, vol. 22, no.8, pp. 951-958, Aug-2003.
[18] S Ravishankar, A Jain, A Mittal, “Automated feature extraction for early detection of diabetic retinopathy in fundus images”,
IEEE Conference on Computer Vision and Pattern Recognition, pp. 210-217, 2009.
[19] M Niemeijer, M D Abramoff, B Van Ginneken, “ Fast detection of the optic disc and fovea in color fundus photographs, Medical
Image Analysis, 13, pp. 859-870, 2009.
[20] M. Niemeijer, J.J. Staal, B. Van Ginneken, M. Loog, M.D. Abramoff,“Comparative study of retinal vessel segmentation methods on
a new publically available database”, SPIE Medical Imaging, 5370, pp. 648656, 2004.

[21] M. Foracchia, E. Grison and A. Ruggeri, “Detection of optic disk in retinal images by means of geometrical mopticdiscel of vessel
structure”, IEEE Trans. Med. Image, vol. 23, no.10, pp. 1189-1195,Oct 2004.
[22] RajuSahebrao Maher, Sangramsing N. Kayte, Sandip T. Meldhe, MuktaDhopeshwarkar, “Automated Diagnosis Non-proliferative
Diabetic Retinopathy in Fundus Images using Support Vector Machine” International Journal of Computer Applications (0975 –
8887)Volume 125 – No.15, September 2015.
[23] RajuSahebrao Maher, Sangramsing N. Kayte, Sandip T. Meldhe, MuktaDhopeshwarkar, “Automated Diagnosis Non-proliferative
Diabetic Retinopathy in Fundus Images using Support Vector Machine” International Journal of Computer Applications (0975 –
8887)Volume 125 – No.15, September 2015.
[24] Geoff Dougherty, “Medical image processing techniques and applications”, Springer.
[25] Deepali A. Gopticdiscse and Dr. Dattatraya S. Bormane, “Automated localization of centre of optic disc and centre of macula in
retinal images”, CiiT International Journal of Biometrics and Bioinformatics, vol. 4, no. 16, pp. 896- 901, Nov 2012.
[26] Bob Zhang and FakhryKarray, “Optic disc detection by multi-scale Gaussian filtering and with scale propticdiscuction and vessels
directional match filter”, Medical Biometrics: Second International Conference, ICMB 2010, Hong Kong, pp. 173-180, 2010.
[27] Seng Soon Lee, MandavaRajeswari and DhaneshRamachandram, “ reliminary and Multi Features Localisation of Optic Disc in
ColourFundus Images”, National Computer Science ostgraduate Colloquium, Malaysia, 2005
[28] Rashid Jalal Qureshi, Laszlo Kovacs, BalazsHarangi, Brigitta Nagy, Tundeeto, Andrasajdu, “Combining algorithms for automatic
detection of optic disc and macula in fundus images”, Elsevier, Computer Vision and Image Understanding, vol. 116, Issue 1, pp.
138146, Jan 2012.

Automated Detection of Optic Disc in Retinal FundusImages Using PCA

More Related Content

What's hot (20)

Similar to Automated Detection of Optic Disc in Retinal FundusImages Using PCA (20)

More from iosrjce (20)

Recently uploaded (20)

Automated Detection of Optic Disc in Retinal FundusImages Using PCA