BLOOD VESSEL ENHANCEMENT AND SEGMENTATION USING WAVELET TRANSFORM
DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING
SREE CHITRA THIRUNAL COLLEGE OF ENGINEERING, THIRUVANANTHAPURAM - 695 018.
Retinal vessel segmentation is an essential step for the diagnoses of various eye diseases. An
automated tool for blood vessel segmentation is useful to eye specialists for purpose of patient screening and clinical study. Vascular
Pattern is normally not visible in retinal images. In this paper, we present a method for
enhancing, locating and segmenting blood vessels in images of retina. We present a method that
uses 2-D Gabor wavelet and sharpening filter to enhance and sharpen the vascular pattern
respectively. This technique locates and segments the blood vessels using edge detection
algorithm and morphological operations. This technique is tested on publicly available STARE
database of manually labeled images which has been established to facilitate
comparative studies on segmentation of blood vessels in retinal images. The validation of our
retinal image vessel segmentation technique is supported by experimental results.
Keywords—Vascular pattern, retinal images, vessel segmentation
BLOOD VESSEL ENHANCEMENT AND SEGMENTATION USING WAVELET TRANSFORM.docx (Size: 1.43 MB / Downloads: 212)
Retinal image processing is greatly required in diagnosing and treatment of many diseases
affecting the retina and the choroid behind it . Diabetic retinopathy is one of the
complications of diabetes mellitus affecting the retina and the choroid. In this condition, a
network of small blood vessels, called choroidal neovascularization (CNV), arises in the
choroid and taking a portion of the blood supplying the retina. As the amount of blood supplying
the retina is decreased, the sight may be degraded and in the severe cases, blindness may occur .
The eye is a window to the retinal vascular system which is uniquely accessible for the study of
a continuous vascular bed in humans. Retinal image processing is greatly required in diagnosing
and treatment of many diseases affecting the retina and the choroid behind it . Diabetic
retinopathy is one of the complications of diabetes mellitus affecting the retina and the choroid.
In this condition, a network of small blood vessels, called choroidal neovascularization (CNV),
arisesin the choroid and taking a portion of the blood supplying the retina. As the amount of
blood supplying the retina is decreased, the sight may be degraded and in the severe cases,
blindness may occur . The eye is a window to theretinal vascular system which is uniquely
accessible for the study of a continuous vascular bed in humans . The detection and
measurement of blood vessels can be used to quantify the severity of disease, as part of the
process of automated diagnosis of disease or in the assessment of the progression of therapy .
Retinal bloodvessels have been shown to have measurable changes in diameter, branching
angles, length , as a result of a disease.
Thus a reliable method of vessel segmentation would be valuable for the early detection and
characterizationofchanges due to such diseases . Retinal vascular pattern facilitates the
physicians for the purposes of diagnosing eye diseases, patient screening, and clinical study .
Inspectionof blood vessels provides the information regarding pathological changes caused by
ocular diseases includingdiabetes, hypertension, stroke and arteriosclerosis . The hand
Retinal vessel segmentation may be used for automatic generation of retinal maps for the
treatment of age-related macular degeneration , extraction of characteristic points of the
retinal vasculature for temporal or multimodal image registration , retinal image mosaic
synthesis,identification of the optic disc position , and localization of the fovea .
The challenges faced in automated vessel detection include wide range of vessel widths, low
contrastwith respect with background and appearance of variety of structures in the image
including the optic disc, the retinal boundary and other pathologies .
Different approaches for automated vessel segmentation have been proposed. Methods based on
vessel tracking to obtain the vasculature structure, along with vessel diameters and branching
points have been proposed by -. Tracking consists of following vessel center
lines guided by local information. In , ridge detection was used to form line elements and
partition the image into patches belonging to each line element. Pixel features were
then generated based on this representation. Many features were presented and a feature
selection scheme is used to select those which provide the best class separability. Papers
- used deformable models for vessels segmentation. Chuadhuri et al.  proposed a
technique using matched filters to emphasize blood vessels. An improved region
based threshold probing of the matched filter response technique was used by Hoover et al. .
In this paper, we present the colored retinal image vessel segmentation technique that enhances
and sharpens the vascular pattern using 2-D Gabor wavelet and sharpening filters.
Our technique creates a binary mask for vessel segmentation applying edge detection algorithm
on sharpened retinal image and a fine segmentation mask is
obtained by applying morphological dilation operation. The paper is organized in four sections.
In Section II, a schematic overview of our implementation methodology is illustrated. Section II
also presents the step by step
techniques required for colored retinal image vessels segmentation. Experimental results of the
tests on the images of the STARE database and their analysis are given in Section III followed by
conclusion in Section IV.
1.2 BLOOD VESSEL ENHANCEMENT AND SEGMENTATION
A schematic overview of proposed blood vessel enhancement and segmentation method is
The monochromatic RGB retinal image is taken as an input and 2-D Gabor wavelet is used to
enhance the vascular pattern especially the thin and less visible vessels are
enhanced using Gabor wavelet . Before extracting vessels from enhanced
retinal image, the blood vessels are sharpened using sharpening filter .
Vessels segmentation binary mask is created by detecting vessels edges from
sharpened image. The blood vessels are marked by the masking procedure which assign
one to all those pixels which belong to blood vessels and zero to non vessels pixels. Final
refined vessel segmentation mask is created by applying morphological dilation operator
Fig. 2 shows the complete flow diagram of proposed blood vessel enhancement and
1.3 EXPERIMENTAL RESULTS
The tests of proposed technique are performed with respect to the vessel segmentation accuracy
and their standard deviation using publicly available STARE database . The STARE database
consists of 20 RGB color images
of the retina. The images are of size 605×700 pixels, 24 bits per pixel (standard RGB). There are
two hand-labeling available for the 20 images of made by two different human observers. The
manually segmented images by 1st human observer are used as ground truth and the
segmentations of set B aretested against set A, serving as a human observer reference for
performance comparison truth. The segmentations of set B are tested against those of A, serving
proposed technique with the accuracies of the methods of Hoover et al. , Staal et al.  and
Soares et al. . Table I summarizes the results of vessel
segmentation for above mentioned methods. It shows the results in term of average accuracy and
their standard deviation. Fig. 6 illustrates the blood vessel segmentation results for proposed
The problem with retinal images is that the visibility of vascular pattern is usually not good. So,
it is necessary toenhance the vascular pattern. In this paper, a wavelet based colored retinal
image blood vessel segmentation technique is proposed. Vascular pattern is enhanced and
edge detection algorithm and morphological dilation operator respectively. We have tested our
technique on publicly available STARE database of manually labeled images. Experimental
results show that our method performs well in enhancing and segmenting the vascular pattern.
Diabetes affects almost one million Australians, and has associated complications such as vision
loss, heart failure and stroke. Any improvement in early diagnosis would therefore represent a
significant gain with respect to reducing the morbidity and mortality of the Australian
population. This work describes one step along the path to providing automated diagnostic
support tools for the early detection of diabetes associated retinal pathology. We focus on the
blood vessel network in the retina that is affected by diabetes. Currently, the
correct assessment of these images using fluorescein requires a specialist, which presents
difficulties in remote or rural areas and is also a relatively invasive and dangerous procedure.
The alternative is to use non-mydriatic colour images that although more difficult to interpret, are
more desirable for use in remote or rural areas as they can be obtained by trained rural health
professionals.The alternative is to use non-mydriatic colour images that although more difficult
to interpret, are more desirable for use in remote or rural areas as they can be obtained by trained
rural health professionals.
Non-mydriatic camera images were analysed in an attempt to fine tune and improve the quality
of retinal blood vessel detection in medical imaging. For this purpose we used the Morlet
wavelet as a tool for segmenting retinal blood vessels in combination with adaptive thresholding,
supervised classification and supervised classifier probabilities combined with
adaptivethresholding. The advantage of wavelet analysis is its multiscale analysing capability in
tuning to specific frequencies, allowing noise filtering and blood vessel enhancement in a
single step. Twenty non-mydriatic camera images of the retinal fundus were analysed, from the
Stare database (http://www.parl.clemson.edu/stare). The results were compared against
the corresponding gold-standard images indicating the true location of vessels as determined by
two ophthalmologists and presented in the manner of free-response receiver operator
characteristic (FROC) curves.
Diabetes is a chronic disease that aﬀects the body’s capacity to regulate the amountof sugar in
the blood. One in twenty Australians are aﬀected by diabetes, but thisﬁgure is conservative, due
to the presence of subclinical diabetes, where the diseaseis undiagnosed, yet is already damaging
the body without manifesting substantial symptoms.
This incidence rate is not conﬁned to Australia, but is typical of de-veloped nations, and even
higher in developing nations. Excess sugar in the bloodresults in metabolites that cause vision
loss, heart failure and stroke, and damage to peripheral blood vessels.These problems contribute
signiﬁcantly to themorbid-ity and mortality of the Australian population, so that any
improvement in earlydiagnosis would therefore represent a signiﬁcant gain. The incidence is
projected to rise, and has already become a major epidemic 
The most common diagnostic test for diabetes is measurement of blood sugar,but this is only
eﬀective when the disease has already made substantial progres-sion. However, because of the
eﬀect of diabetes on peripheral vessels, it is possibleto detect diabetes by examining these
vessels. One of the most suitable areas tomake such an observation is the retina, where small
blood vessels are arrangedon the surface, and visual inspection is possible through the pupil
itself. Thistechnique is well developed, with ophthalmologists routinely employing manual
inspection of the retina for diagnosing diabetic retinopathy, which is causedby diabetes, and
leads to signiﬁcant vision degeneration without prompt treat-ment. In addition cameras can
capture an image of the retina for examination byophthalmologists or for telemedicine as well as
for providing records over time.The requirement of specialists to make an accurate diagnosis
does make retinalphotography prohibitive in cost as a screening tool for the general population
especially in rural or remote regions.
Images containing labelled blood vessels can be derived by injecting a ﬂuorescent dye into the
person being examined, so that blood vessels can be observedwith higher contrast. This
technique, know as ﬂuorescein imaging, is invasiveand brings some risk. As it also requires the
presence of an ophthalmologist, it isnot suitable for rural and remote screening programmes.
Images taken withoutﬂuorescent dye and pupil dilation are known as non-mydriatic, and are also
lessinvasive with good contrast due to the high resolution cameras available. These
are therefore desirable for use in remote or rural areas as they can be obtained
by trained rural health professionals such as indigenous health workers, diabetes
educators and community nurses.The aim of this work is ﬁrst, to improve the accuracy and speed
of vessel segmentation using non-mydriatic retinal images, by the application of advanced
image processing techniques; and second, to apply machine intelligence tech-niques to oﬀer
decision support and reduce the burden on specialist interpretation. Starting with a non-mydriatic
image, our aim is to provide an assessmentof risk of diabetes for the person being examined.
Identiﬁcation of anomalies in retinal blood vessels, associated with diabeteshealth care,
represents a large portion of the assessment carried out by ophthalmologists, which is time
consuming and in many cases does not show anyanomalies at the initial visit. Utilizing non-
specialist health workers in identifying diabetic eye disease is an alternative but trials have
shown that correctidentiﬁcation of retinal pathology may be poor (i.e. only 50% of the cases).
Thissuccess rate decreases for early proliferative retinopathy stages. Telemedicine isan attractive
approach. However, this procedure is not time eﬀective and doesnot lessen the burden on a
comparatively small number of ophthalmologistsin rural areas that need to assess the images. In
addition signiﬁcant technicalproblems lessen the availability of telemedicine
3.1 Image processing for medical diagnosis
Identification of vascular anomalies associated with diabetes health care represents a large
portion of the assessment carried out by ophthalmologists, which is time consuming and in
many cases does not show any anomalies at the initial visit. Utilizing non-specialist health
workers in identifying diabetic eye disease is an alternative but trials have shown that correct
identification of retinal pathology may be poor (i.e. only 50% of the cases). This success rate
decreases for early proliferative retinopathy stages. Telemedicine is an attractive approach.
However, this procedure is not time effective and does not lessen the burden on a
comparatively small number of ophthalmologists in rural areas that need to assess the images. In
addition significant technical problems lessen the availability of telemedicine (Yogesan et 2000).
Automated assessment of blood vessel patterns that can be used by rural health
professionals is now being extended from fluorescein-labelled to non-mydriatic
Automated assessment of blood vessel patterns that can be used by rural healtprofessionals is
now being extended from fluorescein-labelled to non-mydriaticcameraimages (Cesar &Jeline
2003; McQuellin et al., 2002). This work presents the evolution ofretinal blood vessel
segmentation as a function of the effectiveness of the wavelet transform.We outline the use of the
wavelet transform combined with mathematical morphology, supervised training algorithms and
Several methods for segmenting blood vessels using either rule-based or supervised methods
have recently been reported for both fluorescein and non-mydriatic colour retinal
images (Leandro et al., 2003; Sinthanayothin et al., 1999; Staal et al., 2004). Mathematical
morphology, which is a rule-based method, has previously revealed itself as a very useful
digital image processing technique for detecting and counting microaneurysms in fluorescein and
non-mydriatic camera images (Cree et al., 2004; Spencer et al., 1996). Wavelet transformtheory
has grown rapidly since the seminal work by Morlet and Grossman, finding applications in
several realms (Goupillaud et al., 1984). The wavelets space-scale analysis
capability can be used to “decompose” vessel structures into differently scaled Morlet wavelets,
so as to segment them from the retinal fundus.
Automated assessment of blood vessel patterns that can be used by rural health
professionals is now being extended from ﬂuorescein-labelled to non-mydriatic
camera images [3, 15]. This has the advantage of a less invasive and risky proce-
dure, making possible a screening procedure for the general population. A signif-
icant problem in these non-mydriatic images, however, is the ability to identify
the blood vessels in low vessel to background contrast and diverse pathology,
and to separate (segment) them from the background image (fundus). In this
work we present the evolution of retinal blood vessel segmentation, using the
wavelet transform combined with mathematical morphology, supervised train-
ing algorithms and adaptive thresholding. Once the vessels have been success-
fully segmented, it is possible to apply automated measures, such as morphology
measures, then to use further automated methods to identify anomalies.
further processing is outside the scope of this paper, as we concentrate on the
vessel segmentation only.
Several methods for segmenting items of interest have been reported, using
either rule-based or supervised methods for both ﬂuorescein and non-mydriatic
colour retinal images [14, 17, 19]. Mathematical morphology, which is a rule-
based method, has previously revealed itself as a very useful digital image pro-
cessing technique for detecting and counting microaneurysms in ﬂuorescein and
non-mydriatic camera images [4, 12, 18]. Wavelet transform theory has grown
rapidly since the seminal work by Morlet and Grossman, ﬁnding applications in
many realms (e.g. ). The wavelets space-scale analysis capability can be used
to decompose vessel structures into diﬀerently scaled Morlet wavelets, so as to
segment them from the retinal fundus.
The recognition of images, or parts of images as possessing pathologies, has
responded well to automated classiﬁcation techniques. Here the key is to deter-
mine some relationship between a set of input vectors that represent stimuli, and
a corresponding set of values on a nominal scale that represent category or class.
The relationship is obtained by applying an algorithm to training samples that
are 2-tuples (u,z), consisting of an input vector u and a class label z. The learned
relationship can then be applied to instances of u not included in the training
set, in order to discover the corresponding class label z
The learned relationship can then be applied to instances of u not included in the training
set, in order to discover the corresponding class label z . This process, known
as supervised classiﬁcation, requires manually labelled images for training the
model, and also requires suitable measures to form the vector u. These measures
can be derived from the previously discussed techniques, including mathematical
morphology and the wavelet transform. After training, the model can then be
used to classify previously unseen images. Alternatively, it is possible to classify
individual pixels as either belonging to a vessel or to the background of the im-
age. The classiﬁcation technique can include Artiﬁcial Neural Networks or many
others from the range of techniques available.
In methods 2 and 3, supervised classification was applied to obtain the final segmentation,
with the pixel classes defined as C1 = vessel-pixels and C2 = non-vessel pixels using the
Bayesian classifier consisting of a mixture of Gaussians (Theodoridis, 1999). In order to obtain
the training set, retinal fundus images have been manually segmented, thus allowing the creation
of a labelled training set into 2 classes C1 and C2 (i.e. vessels and non-vessels).
In this work, the hand-drawn vascular tree provided by the ophthalmologist has been used – our
training pattern – so that we obtained its feature space. Two different strategies for deriving the
training set were applied:
1. Some images were completely segmented by an expert and a random sub-set of their
pixelswas used to train the classifier.
2. Only a small portion (window) of a sample image was manually segmented. The labelled
pixels are then used to train the classifier, which is applied to the same image in order to
conclude its segmentation.
3. This 2ndstrategy has been devised so that a semi-automated fundus segmentation
software can be developed, in which the operator only has to draw a small portion of the
vessels over the input image or simply click on several pixels associated with the vessels
The results were obtained in the form of segmented images and compared by the experts. Here
we present the results in the manner of FROC curves to assess the trade off between true-
positive fraction and false-positive fraction of our methods in detecting the pixels associated
with the vessel patterns.
In this work we assess the relative merits of several techniques for segmentation
of blood vessels from colour retinal images. Twenty digital images were used from
the Stare database . This database also includes the opinions of two experts
who had indicated the position of the vessels from colour images to establish
two “gold standards” as separate images.Our strategy was to use three methods for segmenting
retinal blood vesselsfrom directly digitized colour retinal images. The experimental procedure
followed was to pre-process the images ﬁrst to optimise the use of the wavelet
transforms. The methods tested were:
1. Wavelet transform plus adaptive thresholding,
2. Wavelet transform plus supervised classiﬁers,
3. Wavelet transform plus pixel probabilities combined with adaptivethresholding.
In addition, we compared two training techniques: training on one or more
complete images, then classifying the remaining images, and training on a win-
dow of the image then classifying the remainder of the same image.
Initially the methods were compared qualitatively, but the best of these methods were selected
and compared numerically by plotting on a graph of true positive against false positive results
from the classiﬁcation. This graph resembles
a free-response receiver operating characteristic (FROC) curve to aid the readerin its
True positives occur when the classiﬁer labels a pixel as
belonging to a vessel and the gold standard segmentation also labels the pixelas vessel.
In order to reduce the noise eﬀects associated with the processing, the inputimage was pre-
processed by a mean ﬁlter of size 5×5 pixels. Due to the circularshape of the non-mydriatic
image boundary, neither the pixels outside the region of-interest nor its boundary were
considered, in order to avoid boundary eﬀects.For our wavelet analysis we used the green
channel of the RGB components ofthe colour image as it displayed the best vessels/background
In order to reduce the noise effects associated with the processing, the input image was pre –
processed by a mean filter of size 5x5 pixels. Due to the circular shape of the non-mydriatic
image boundary, neither the pixels outside the region-of-interest nor its boundary were
considered, in order to avoid boundary effects. For our wavelet analysis we used the green
channel of the RGB components of the colour image as it displayed the best
6.2 Continuous wavelet transform plus adaptive thresholding
Applying the continuous wavelet transform approach provides several benefits but resulted in
some loss of detail as the scale parameter was fixed. We therefore adopted a pixel
thresholding approach that represented each pixel by a feature vector including colour
information, measurements at different scales taken from the continuous wavelet (Morlet)
transform and the Gaussian Gradient, as well as from mean filtering applied to the green
channel. The resulting feature space was used to provide an adaptive local threshold to assign
each pixel as either a vessel-pixel or a non-vessel pixel.
The real plane RR × is denoted as
2R , and vectors are represented as bold letters, e.g.
2, R ∈ b x . Let
2L f ∈ be an image represented as a square integrable (i.e. finite energy)
function defined over 2
R (Arnéodo et al., 2000). The continuous wavelet transform (CWT) is
Applying the continuous wavelet transform approach provides several beneﬁts
but resulted in some loss of detail as the scale parameter was ﬁxed. We therefore
adopted a pixel thresholding approach that represented each pixel by a feature
vector including colour information, measurements at diﬀerent scales taken from
the continuous wavelet (Morlet) transform and the Gaussian Gradient, as well
as from mean ﬁltering applied to the green channel. The resulting feature space
was used to provide an adaptive local threshold to assign each pixel as either a
vessel-pixel or a non-vessel pixel.
7. Feature Extraction
The pixel feature space was formed by Morlet wavelet responses (taken at dif-
ferent scales and elongations), Gaussian Gradient responses (taken at diﬀerent
scales) and colour information, which determine each pixel’s colour. This re-
sulted in a computationally demanding high dimensional feature space. At the
same time, Morlet responses taken at close scales are highly correlated, as are
the Gaussian Gradient responses for similar scales. Therefore we used a feature
extraction approach to obtain a lower dimensional feature space, while trying
to preserve structure important for discrimination. Feature extraction was per-
formed by a linear mapping provided by nonparametric discriminant analysis
. Nonparametric discriminant analysis consists of building two matrices. The
ﬁrst is a nonparametric between-class scatter matrix, constructed using k-nearest
neighbour techniques, which deﬁnes the directions of class separability. The sec-
ond is the within-class scatter matrix, which shows the scatter of samples around
their mean class vectors. These matrices were built based on the labelled training
samples. The two matrices are then used to ﬁnd a projection (given by a linear
mapping) that maximizes class separability while minimizing the within-class
scatter in the projected feature space.
We have used 2-D Gabor wavelet to enhance the
vascular pattern and thin vessels . 2-D Gabor wavelet is used due to its directional
selectiveness capability of detecting oriented features and fine tuning to specific frequencies
Fig. 3 shows the enhanced vascular pattern afterapplying Gabor wavelet. Fig. 3(b) shows the
enhancedvascular pattern using Hoover et al. method  while fig.
3© shows the vascular pattern enhanced using waveletmethod and there is a clear improvement
inenhancementusing wavelet method.
7.2 Sharpening Filter
We used unsharp masking  on enhanced vessel retinal image to sharpen the vascular pattern
. The application of Gabor wavelet on colored retinal image enhances the vascular pattern but the
resulting image is a
little blurred so we have used unsharp filter to sharpen the vascular edges. This helps in reliable
extraction of vessels from the colored retinal image. Fig. 4(a) shows enhanced retinal image and
fig. 4(b) shows sharpened retinal image. It is clearly visible that the vessels are much more
prominentthan they were in the original image.
7.3Vessel Segmentation Mask
Vessels segmentation mask is created by extracting vessels boundaries using edge detection
algorithm and then
by applying morphological dilation operator . In this paper, we have used canny edge
detector . Canny operator simultaneously optimizes three criteria: detection criterion,
localization criterion, and elimination of multiple responses and these rules also become a
standard to evaluate edge detection algorithm performance . The vessels extracted using edge
detection algorithm are
refined using standard morphological dilation operator . Dilation will fill the gaps between
vessel boundaries detected by edge detection algorithm.
7.4 MORPHOLOGICAL DIALATION OPERATION
Before we present the results of the experiments comparing the different applications of the
segmentation procedures, we provide for comparison an example of the application of the
wavelet transform to fluorescein images (Leandro et al., 2001). Figure 1(a) shows a typical
image of the retinal fundus with the optic disk on the right hand side and the protruding blood
vessels that course throughout the image. Figure 1 (b) shows the result of image segmentation
using the Morlet wavelet transform. The latter shows the difficulty in obtaining a clear
segmentation. Background noise and variable grey levels across the image introduce artefacts. In
particular, this method did not remove the optic disc located on the right hand side of the image
and was very susceptible to hue variation that resulted in the large grey area to the left of the
We have demonstrated some new techniques for the automated processing of non-mydriatic
images in the study of diabetic retinopathy that can certainly be extended to other contexts in
pattern recognition. The results we have obtained so far suggest that pixel classification, in
conjunction with wavelet transform and adaptive thresholding, can provide noise-robust vessel
segmentation. The approach reported here improved on previous results by reducing
the level of interaction required with the segmentation program, providing a useful tool for non-
specialists such as community health workers in assessing fundus complications associated with
diabetes (Antoine et al., 1997; Cesar &Jelinek, 2003; da Costa, 2001; Grossmann, 1988; Zana&
Wavelets are especially suitable for detecting singularities (e.g. edges) in signals,
extracting instantaneous frequencies, and performing fractal and multifractal analysis (Antoine et
al., 1997; Grossmann, 1988). Applying the wavelet transform allows noise filtering and blood
vessel enhancement in a single step. Our results indicate that for the same false-positive fraction,
the supervised learning with adaptive thresholding obtained a greater than 75% sensitivity
compared to the ophthalmologist with approximately 90%. Future research is now focused on
fine tuning our algorithms on a larger data set.
The problem with retinal images is that the visibility of vascular pattern is usually not good. So,
it is necessary to enhance the vascular pattern. In this paper, a wavelet based
colored retinal image blood vessel segmentation technique is proposed. Vascular pattern is
enhanced and sharpened prior to their detection. Vessel and located and segmentation
mask is created using edge detection algorithm and morphological dilation operator
respectively. We have
tested our technique on publicly available STARE database of manually labeled images.
Experimental results show that our method performs well in enhancing and segmenting the
M. S. Mabrouk, N. H. Solouma and Y. M. Kadah,
“Survey of retinal image segmentation and registration” , GVIP Journal, vol. 6, No.2,
 M. E. Martnez-Prez, A. D. Hughes, S. A. Thom, A.
A. Bharath and K. H. Parker, “Retinal Blood Vessel Segmentation by Means of Scale-Space
C. Analysis and Region Growing ”, Medical Image Computing and Computer-Assisted
E. Intervention MICCAI99, pp. 90-97, 1999.
 J. KansKy, Clinical Opthalmology, Butterworh-
Heinmann , London, 1994.
 T. Teng, M. Lefley and D. Claremont, “Progress towards automated diabeticocular
screening: A review of image analysis and intelligent systems
for diabetic retinopathy,” Med. Biol. Eng. Comput., vol. 40, pp. 2-13, 2002.
 E. J. Susman, W. J. Tsiaras, and K. A. Soper,
“Diagnosis of diabetic eye disease,” JAMA, vol. 247, pp. 3231-3134, 1982.
 A. Pinz, S. Bernogger, P. Datlinger and A. Kruger,
“Mapping the human retina,” IEEE Trans. Med. Imag., vol. 17, no. 4, pp. 606-619, Aug. 1998.
 C.L. Tsai, C. V. Stewart, H. L. Tanenbaum and B. Roysam, “Modelbased method for
improving the accuracy and repeatability of estimating vascular bifurcations and crossovers from