Conference PaperPDF Available

An Automated Region-Of-Interest Segmentation for Optic Disc Extraction

Authors:

Abstract and Figures

Optic disc segmentation in retinal fundus images plays a critical rule in diagnosing a variety of pathologies and abnormalities related to eye retina. Most of the abnormalities that are related to optic disc lead to structural changes in the inner and outer zones of optic disc. Optic disc segmentation on the level of whole retina image degrades the detection sensitivity for these zones. In this paper, we present an automated technique for the Region-Of-Interest segmentation of optic disc region in retinal images. Our segmentation technique reduces the processing area required for optic disc segmentation techniques leading to notable performance enhancement and reducing the amount of the required computational cost for each retinal image. DRIVE, DRISHTI-GS and DiaRetDB1 datasets were used to test and validate our proposed pre-processing technique.
Content may be subject to copyright.
An Automated Region-Of-Interest Segmentation for
Optic Disc Extraction
Jasem Almotiri and Khaled Elleithy
Computer Science Department
University of Bridgeport
Bridgeport, CT 06604 USA
jalmotir@my.bridgeport.edu, elleithy@bridgeport.edu
Abdelrahman Elleithy
Computer Science Department
William Paterson University
Wayne, NJ 07470, USA
ElleithyA@wpunj.edu
Abstract—Optic disc segmentation in retinal fundus images plays
a critical rule in diagnosing a variety of pathologies and
abnormalities related to eye retina. Most of the abnormalities that
are related to optic disc lead to structural changes in the inner and
outer zones of optic disc. Optic disc segmentation on the level of
whole retina image degrades the detection sensitivity for these
zones. In this paper, we present an automated technique for the
Region-Of-Interest segmentation of optic disc region in retinal
images. Our segmentation technique reduces the processing area
required for optic disc segmentation techniques leading to notable
performance enhancement and reducing the amount of the
required computational cost for each retinal image. DRIVE,
DRISHTI-GS and DiaRetDB1 datasets were used to test and
validate our proposed pre-processing technique.
Keywords-Region-Of-Intreset; optic disc; optic disc
segmentation; retina; Hough transform; fuzzy c-means.
I. INTRODUCTION
Retina fundus images are used by ophthalmologists in the
diagnosis of eye-related diseases. The analysis stage of retina
image that follow the image capturing is considered a corner
stone in the overall diagnosis process. Retinal image analysis
includes locating and extracting many retinal anatomical
structures in a separate view which eases the diagnosis, gives
more insight and thereby, enhances the diagnosis accuracy.
Retinal vessels, Fovea, different types of lesions, optic disc
shown in Figure 1 are typical examples of retinal anatomical
structures that represent target objects for a broad range of
segmentation techniques and algorithms [1-4].
All methods designed for retinal image analysis or other
objects consist of three major stages; Pre-Processing,
Processing and Post-Processing. Pre-processing stage is
considered a preliminary one since it affects the quality of
subsequent stages and affects the overall segmentation
performance [5]. This stage includes one or more of the
operations; noise/artifacts removal, green, or red layer
extraction, image contrast enhancement, and Region-Of-Interest
(ROI) extraction. However, some methods used to reduce high
dimensional data in the pre-processing [6].
An automated ROI detection module was proposed by Zhang
et al. [7] using an image pre-processing step called fringe
removal . The authors observed that the unbalanced brightness
found in the fringe of the optic disc rim goes behind failure in
the optic disc.
Dehghani et al. [8] used the average of the histogram to
localize the optic disc regions while other researchers exploited
the elliptic shape of the optic disc and wavelet transform for
performing this task as in [9]. Average filtering process
followed by thresholding was used by [10] whereas, an iterative
thresholding process followed by connected component
analysis was used by [11] for optic disc center localization as a
pre-step before optic disc contour localization.
Instead of letting the user to manually choose the optic disc
region, this paper presents an intelligent automated ROI
localization, and extraction technique based on fuzzy c-means
and Hough transform. The rest of this paper is organized as
follows. Section II presents the proposed region-of-interest
extraction scheme. Section III presents experimental results, and
Section IV concludes the paper.
II. PROPOSED METHOD
In this work, we propose a new technique that combines two
techniques; Fuzzy c-means and Hough transform. Fuzzy c-
means associated with morphological operations was used as an
edge-map creator whereas Hough transform was utilized for the
optic disc center localization. Morphological operations played
more than a complement step; namely, it plays a central role in
case of extracting the optic disc ROI in pathological fundus
images. The general flowchart of the proposed technique is
Figure 1. Retinal vessels and optic disc anatomical structures.
illustrated in Figure 2.
Our identification technique is composed of the following
four major stages: (1) Retinal image pre-processing. (2) Fuzzy
c-means binarization. (3) Optic disc localization. (4) Optic disc
cropping.
A. Retinal Image Pre-processing
Optic disc region normally exhibits a high contrast in
comparison to other anatomical structures in digital fundus
images. However, the existence of large vessels passing through
the optic disc region besides possible neighbor lesions makes
the image contrast enhancement a very first step. Since the
contrast of the optic disc in the red channel layer is better in
comparison to blue and green layers of RGB color fundus
image; we extracted the red layer of raw retina image then the
successive steps of pre-processing were applied on it as shown
in Figure 3.a. In the second step of the pre-processing stage, a
Contrast Limited Adaptive Histogram Equalization (CLAHE)
followed by a median filtering of 9×9 sized-window were
applied on the gray scale image found in first step of this stage,
as illustrated in Figure 3.b and 3.c.
B. Fuzzy c-means Binarization
Optic disc appears as circular or semi-elliptical spot on the
surface of retina. Once the optic disc has been localized; ROI
extraction proceeds smoothly. Thus, we utilized Hough
transform to localize the center of the optic disc circle.
However, Since Hough transform was originally designed for
lines and parametric curves detection, edge detection is often
(a) (b) (c)
Figure 3. Pre-processing steps: (a) Red layer output. (b) CLAHE output. (c) Median filter output.
Figure 2. Major stages of optic disc ROI extraction technique.
used as pre-processing step to Hough transform. Therefore, we
use fuzzy c-means [12] for the sake of edge-map detection.
Using c = 25, the pre-processed image was clustered into 25
clusters (image of 25 gray levels. The number of clusters was
chosen, in such a way that one cluster (of high gray level = c)
corresponds to the optic disc region or at least part of it, and the
other corresponds to the other background objects as illustrated
in Figure 4.b The output of fuzzy c-means algorithm is
binarized as in (1):

=
1;

=
0,ℎ. (1)
Where

is the binarized version of fuzzy c-means
output

.
C. Optic Disc Localization
In order for Hough transform to work, it requires an edge-
map image. Moreover, the binarized image of the fuzzy c-
means binarization may contain residual white spots
corresponding to fundus camera artifacts, noise or as often due
to neighbor retina lesions such as hard and soft exudates lesions.
Therefore, as a pre-processing step before using Hough
transform for optic disc center localization, we carried out a set
of successive morphological operations in order to remove
these artifacts and create the edge-map image. Morphological
opening followed by dilation removes the objects (of size less
than or equals 1500 pixels). Then, the boundary pixels (edge-
map) of clean region is established as elaborated in Figure 5.
(b), (c) and (d).
D. Optic Disc Cropping
Since Hough transform detects the coordinates
(

,

) of the optic disc circle as shown in Figure 5.
(e), a perfect circle can be synthesized with a radius . Choosing
the radius value depends on the validation dataset used; because
each dataset produced using fundus camera with particular
specifications in terms of image size and pixel resolution.
Radius value was used in our technique to establish the
widows’ borders of the optic disc region as in (2):

ℎ
=2∗

ℎ
=2∗
(2)
Then, using dimensions specified in (2) and MATLB ®
image cropping function, the final optic disc ROI has been
extracted as shown in Figure 5. (f).
III. EXPERIMENTAL RESULTS AND DISCUSSION
This section explores and discusses the experimental results
obtained as a result of running our proposed optic disc ROI
extraction system. The proposed system was developed and
tested in the environment of MATLAB 2017a using image
processing toolbox, using a personal laptop configured with
Windows 10 professional with Intel Core i7 4Due 2.4,1.8GHZ
CPU and 8.0 GB RAM.
In this section, the success of our proposed system in
identifying and extracting the optic disc region in a non-
pathological retina images was evaluated using public
DRISHTI-GS dataset [13], which contains retinal fundus
images for 50 patients using a 30 degrees field of view (FOV)
at a resolution of 2896 x 1944 pixels. Figure 6 shows typical
examples of optic disc localization and ROI extraction results
of our proposed technique applied on DRISHTI-GS dataset
where is 400 pixels.
Figure 4. Fuzzy c-means output of a pathological retina image. (a) Actua
l
fuzzy c-means output of c = 25 clusters yields 25 gray levels graduall
y
decrease as we move away of the optic disc center. (b) color-substituted image
corresponds to image in (a) for more gray levels clarification. (c) Binarized
version of fuzzy c-means output where it shows a part of optic disc center
(large white spot), and other small spots represents exudates lesions in this
image.
As can be noted from Figure 6, the proposed technique is able
to localize and extract the ROI successfully with the presence
of camera artifacts and noise. This is due to applying fuzzy c-
means as an edge-map tool for Hough transform rather than
directly using a thresholding technique.
For pathological fundus images, we have validated our
technique via DiaRetDB1 [14] dataset at the image level. This
dataset consists of 89 color fundus images taken in Kuopio
University Hospital of diabetic retinopathy abnormalities.
Other images are normal, according to four experts involved in
the diagnostic process. Figure 7 illustrates typical examples of
the optic disc localization as well as the corresponding ROI
extraction results of our proposed technique based on the
pathological DiaRetDB1 dataset. As shown in Figure 7, the
proposed technique is able to detect the optic disc region in
pathological fundus images due to high clustering capability of
the fuzzy c-means without the need for texture analysis.
For the sake of results comparison, an optic disc localization
matched up to 96% of cases was reported by [7] segmentation
results using ORIGA dataset [15], where the rest of images
(remaining of 4%) have bad quality. In these cases, the user
need to adjust OD ROI manually. However, our proposed
technique showed better results in terms of localization and OD
ROI segmentation where OD localization reached up to 100%
for normal (and semi-normal) retina fundus images based on
both DRISHTI-GS and DRIVE datasets whereas it has achieved
78.38% for subset of 37 pathological fundus images chosen
from DiaRetDB1 dataset.
These results support the validity of our ROI segmentation
technique and show that our technique gives exceptional results
for both low and high noisy and pathological areas.
IV. CONCLUSION
One of the challenging issues with different types of
segmentation tasks applied on retinal images is that the quality
of acquired images is usually not good. This is due to noise or
camera artifacts including uneven illumination, blurred and
noisy regions as well as different types of retina pathologies. It
is of considerable importance to enhance the quality of colored
retinal images for reliable detection or segmentation. In this
paper, an ROI extraction pre-processing technique was
proposed for different types of retinal image segmentation. It
was validated on both normal and abnormal retina images
where it showed supreme performance in comparison to other
optic disc ROI extraction techniques.
Figure 5. Optic disc localization and cropping: (a) abnormal (pathological)
fundus image. (b) Fuzzy c-means binarization output. (c) Morphological
cleaning. (d) Edge-map detection. (e) Optic disc center localization. (d) Optic
disc ROI window.
Figure 6. Optic disc localization and ROI segmentation results for the DRISHTI-DS dataset: Column 1: Original retinal images.
Column 2: OD localization Column 3: Corresponding OD region of interest.
Figure 7. Optic disc localization and ROI segmentation results for the pathological DiaRetDB1 dataset. Column 1: Original retinal
images. Column 2: OD localization Column 3: Corresponding OD region of interest.
R
EFERENCES
[1] J. Almotiri, K. Elleithy, and A. Elleithy, "Retinal Vessels Segmentation
Techniques and Algorithms: A Survey," Applied Sciences, vol. 8, p. 155,
2018.
[2] C. L. Srinidhi, P. Aparna, and J. Rajan, "Recent advancements in retinal
vessel segmentation," Journal of medical systems, vol. 41, p. 70, 2017.
[3] A. Almazroa, R. Burman, K. Raahemifar, and V. Lakshminarayanan,
"Optic disc and optic cup segmentation methodologies for glaucoma
image detection: a survey," Journal of ophthalmology, vol. 2015, 2015.
[4] M. M. Fraz, P. Remagnino, A. Hoppe, B. Uyyanonvara, A. R. Rudnicka,
C. G. Owen, et al., "Blood vessel segmentation methodologies in retinal
images–a survey," Computer methods and programs in biomedicine, vol.
108, pp. 407-433, 2012.
[5] M. Daneshzand, R. A. Zoroofi, and M. Faezipour, "MR image assisted
drug delivery in respiratory tract and trachea tissues based on an enhanced
level set method," in Proceedings of the 2014 Zone 1 Conference of the
American Society for Engineering Education, 2014, pp. 1-7.
[6] J. Almotiri, K. Elleithy, and A. Elleithy, "Comparison of autoencoder and
Principal Component Analysis followed by neural network for e-learning
using handwritten recognition," in 2017 IEEE Long Island Systems,
Applications and Technology Conference (LISAT), 2017, pp. 1-5.
[7] Z. Zhang, B. H. Lee, J. Liu, D. W. K. Wong, N. M. Tan, J. H. Lim, et al.,
"Optic disc region of interest localization in fundus image for glaucoma
detection in ARGALI," in Industrial Electronics and Applications
(ICIEA), 2010 the 5th IEEE Conference on, 2010, pp. 1686-1689.
[8] A. Dehghani, H. A. Moghaddam, and M.-S. Moin, "Optic disc localization
in retinal images using histogram matching," EURASIP Journal on Image
and Video Processing, vol. 2012, p. 19, 2012.
[9] P. Pallawala, W. Hsu, M. L. Lee, and K.-G. A. Eong, "Automated optic
disc localization and contour detection using ellipse fitting and wavelet
transform," in European conference on computer vision, 2004, pp. 139-
151.
[10] M. U. Akram, A. Khan, K. Iqbal, and W. H. Bu tt, "Retinal images: optic
disk localization and detection," in International Conference Image
Analysis and Recognition, 2010, pp. 40-49.
[11] P. Siddalingaswamy and K. G. Prabhu, "Automatic localization and
boundary detection of optic disc using implicit active contours,"
International Journal of Computer Applications, vol. 1, pp. 1-5, 2010.
[12] J. C. Bezdek, R. Ehrlich, and W. Full, "FCM: The fuzzy c-means
clustering algorithm," Computers & Geosciences, vol. 10, pp. 191-203,
1984.
[13] J. Sivaswamy, S. Krishnadas, G. D. Joshi, M. Jain, and A. U. S. Tabish,
"Drishti-gs: Retinal image dataset for optic nerve head (onh)
segmentation," in Biomedical Imaging (ISBI), 2014 IEEE 11th
International Symposium on, 2014, pp. 53-56.
[14] T. Kauppi, V. Kalesnykiene, J.-K. Kamarainen, L. Lensu, I. Sorri, A.
Raninen, et al., "The DIARETDB1 Diabetic Retinopathy Database and
Evaluation Protocol," in BMVC, 2007, pp. 1-10.
[15] Z. Zhang, F. S. Yin, J. Liu, W. K. Wong, N. M. Tan, B. H. Lee, et al.,
"Origa-light: An online retinal fundus image database for glaucoma
analysis and research," in Engineering in Medicine and Biology Society
(EMBC), 2010 Annual International Conference of the IEEE, 2010, pp.
3065-3068.
... In addition to that adjacent pixels that are corresponding to the given region of an image are selected, based on definite attribute like color similarity, intensity range and so on. In this algorithm, the input image can be separated into '2' regions such as foreground and background [9]. It requires an easy mode to do so is to select a seed 'XϵI' and then enlarges it. ...
... The first phase of VST is preprocessing of an image is accepted with contrast improvement by brightening the object from the background [9], hence that the 'ROI' can simply be distinguished from the fundus eye image. In proposed scheme, contrast improvement with 'Histogram Equalization'(HE) handling is working on the frequency of input FE image after noise removal. ...
Article
In medical image processing, segmentation is the one of the significant area of clinical diagnosis and image processing phases. In’retinal fundus’ images, optic disc segmentation plays an important role in analyzing several pathologies, including the abnormalities associated to’eye retina’. Emerging the perfect automated segmentation of glaucoma region is an extremely hard because of many anomalies present in the input image. In this research paper, presents a method for segmentation/extraction of’Optic Disc (OD) region’ from retinal images along with ROI (Region of Interest) is generated automatically by Visual Saliency Thresholding(VST) method. The mathematical morphology is fully utilized in the proposed segmentation technique. The ROI of OD is recognized by optic cup region with threshold value by means of markers, whereas optic disc is extracted from visual regions through saliency parameters. The achieved experimental results are positive. For assessing the proposed segmentation method that has been verified with many images also compared with the other segmentation algorithms such as Otsu thresholding and region growing. The implementation results direct that proposed technique can be achieved the maximum accuracy than other algorithms. DRISHTI-GS1 dataset is utilized to approve the performance of our proposed segmentation system.
... Most of the abnormalities related to the optical disc will cause structural changes in the internal and external areas of the optical disc. The segmentation of the optic disc at the image level of the entire retina increases the detection sensitivity of these regions [13]. ...
Chapter
Full-text available
Glaucoma is a disease that affects the vision of those who suffer from it, leading to irreversible vision loss. The cup-to-disc relationship is an indispensable landmark for diagnosing glaucoma; therefore, performing an accurate disc and optic cup segmentation is desirable. In this work, a methodology to separately segment the optic disc and the optic cup from color fundus retina images is presented. For this purpose, the cup and disk sections were manually segmented from the public ORIGA database with the help and of specialists in the field of glaucoma. The complete retina images are cropped, identifying the region of interest (ROI), then converted to polar coordinates to minimize imbalance between classes. Then, an encoder-decoder architecture is applied with residual blocks for segmentation, obtaining an accuracy of 96.94% for the optic disc and 93.2% for the optic cup, as well as an overlap error of 0.088 and 0.0706, respectively.
... Therefore the optic disc region is typically removed to examine and classify the vessel structure. In this proposed methodology OD is detected using Fuzzy c-means clustering [22] and removed and mentioned in fig. 6. ...
Chapter
Deep convolution neural networks (DCNNs) are playing critical roles in various computer vision tasks, though they also suffer from many problems. The performance of DCNNs will severely degrade when facing gaps between the training set and test set. It is quite common in the field of medical image analysis where domain shift between images acquired from different devices or patients. A feasible solution to this problem is collecting more labeled data from the testing set for fine-tuning DCNNs to improve the generalization ability. Unfortunately, medical image labeling is complicated and time-consuming, which makes fine-tuning hard to apply. In this paper, we propose a novel network model which makes full use of the existing pre-trained network models, effectively reducing domain shift and restraining performance degradation. This method dramatically reduces the dependence on labeled data and achieve better results even if trained with a small amount of unlabeled data. We transform the testing images (target domain) with adversarial learning mechanism and make them look similar to training data (source domain) that can be segmented directly by the pre-training model trained on the training set. In addition, we introduce additional structural consistency constraints to suppress the distortion during forwarding propagation when the model is trained with fewer samples, which ensure the structure of the generated image is consistent with the input image. We validate our method in retinal fundus image segmentation task. Experiments show that the proposed method suppresses the degradation of model performance caused by domain migration, and achieves almost the same segmentation performance as the original training data.
Article
Full-text available
Retinal vessels identification and localization aim to separate the different retinal vasculature structure tissues, either wide or narrow ones, from the fundus image background and other retinal anatomical structures such as optic disc, macula, and abnormal lesions. Retinal vessels identification studies are attracting more and more attention in recent years due to non-invasive fundus imaging and the crucial information contained in vasculature structure which is helpful for the detection and diagnosis of a variety of retinal pathologies included but not limited to: Diabetic Retinopathy (DR), glaucoma, hypertension, and Age-related Macular Degeneration (AMD). With the development of almost two decades, the innovative approaches applying computer-aided techniques for segmenting retinal vessels are becoming more and more crucial and coming closer to routine clinical applications. The purpose of this paper is to provide a comprehensive overview for retinal vessels segmentation techniques. Firstly, a brief introduction to retinal fundus photography and imaging modalities of retinal images is given. Then, the preprocessing operations and the state of the art methods of retinal vessels identification are introduced. Moreover, the evaluation and validation of the results of retinal vessels segmentation are discussed. Finally, an objective assessment is presented and future developments and trends are addressed for retinal vessels identification techniques.
Article
Full-text available
Retinal vessel segmentation is a key step towards the accurate visualization, diagnosis, early treatment and surgery planning of ocular diseases. For the last two decades, a tremendous amount of research has been dedicated in developing automated methods for segmentation of blood vessels from retinal fundus images. Despite the fact, segmentation of retinal vessels still remains a challenging task due to the presence of abnormalities, varying size and shape of the vessels, non-uniform illumination and anatomical variability between subjects. In this paper, we carry out a systematic review of the most recent advancements in retinal vessel segmentation methods published in last five years. The objectives of this study are as follows: first, we discuss the most crucial preprocessing steps that are involved in accurate segmentation of vessels. Second, we review most recent state-of-the-art retinal vessel segmentation techniques which are classified into different categories based on their main principle. Third, we quantitatively analyse these methods in terms of its sensitivity, specificity, accuracy, area under the curve and discuss newly introduced performance metrics in current literature. Fourth, we discuss the advantages and limitations of the existing segmentation techniques. Finally, we provide an insight into active problems and possible future directions towards building successful computer-aided diagnostic system.
Article
Full-text available
Glaucoma is the second leading cause of loss of vision in the world. Examining the head of optic nerve (cup-to-disc ratio) is very important for diagnosing glaucoma and for patient monitoring after diagnosis. Images of optic disc and optic cup are acquired by fundus camera as well as Optical Coherence Tomography. The optic disc and optic cup segmentation techniques are used to isolate the relevant parts of the retinal image and to calculate the cup-to-disc ratio. The main objective of this paper is to review segmentation methodologies and techniques for the disc and cup boundaries which are utilized to calculate the disc and cup geometrical parameters automatically and accurately to help the professionals in the glaucoma to have a wide view and more details about the optic nerve head structure using retinal fundus images. We provide a brief description of each technique, highlighting its classification and performance metrics. The current and future research directions are summarized and discussed.
Conference Paper
Full-text available
Optic nerve head (ONH) segmentation problem has been of interest for automated glaucoma assessment. Although various segmentation methods have been proposed in the recent past, it is difficult to evaluate and compare the performance of individual methods due to a lack of a benchmark dataset. The problem of segmentation involves segmentation of optic disk and cup region within ONH region. Available datasets do not incorporate challenges present in this problem. In this data paper, we present a comprehensive dataset of retinal images which include both normal and glaucomatous eyes and manual segmentations from multiple human experts. Both area and boundary-based evaluation measures are presented to evaluate a method on various aspects relevant to the problem of glaucoma assessment.
Conference Paper
Full-text available
In medical diagnosis and therapy, finding an appropriate method to evaluate the effect of various drugs is crucial. There are several ways to qualify a drug for a specific disease and one way is through medical image analysis. This process varies with the tissues we want to analyze and the imaging technique that is employed. For hydrous tissues such as nasal and trachea, Magnetic Resonance Imaging can be helpful for further evaluations. Trachea can be challenged by an antigen which will increase both nasal vascular permeability and intranasal pressure. Another effect of antigen challenge into nasal cavity which may cause nasal blockage, is swelling of nasal mucosa and a decrease in nasopharyngeal airway. In this paper, we study the effect of an antihistamine drug on swelling of mucosa. This antihistamine is called Azelastine and is injected to guinea pig to evaluate the swelling changes of nasal and trachea mucosa. After 20 minutes of injection, a MR image of the motionless animal is taken and this imaging will continue for 30, 40, 50 and 60 minutes from injection. Due to the ambiguous nature of respiratory tract, finding a precise method for processing has useful results. Watershed algorithm has widespread function in medical images but its defects in segmentation can be modified by different methods. An enhanced level set method is used here; a nonparametric active contour for nasal and trachea detection. This automatic image segmentation and tissue detection can help physicians evaluate the effect of a specific drug from medical images.
Article
Full-text available
In this article, we propose a new method for localizing optic disc in retinal images. Localizing the optic disc and its center is the first step of most vessel segmentation, disease diagnostic, and retinal recognition algorithms. We use optic disc of the first four retinal images in DRIVE dataset to extract the histograms of each color component. Then, we calculate the average of histograms for each color as template for localizing the center of optic disc. The DRIVE, STARE, and a local dataset including 273 retinal images are used to evaluate the proposed algorithm. The success rate was 100, 91.36, and 98.9%, respectively.
Conference Paper
Full-text available
Glaucoma is the second leading cause of permanent blindness worldwide. Glaucoma can be diagnosed through measurement of neuro-retinal optic cup-to-disc ratio (CDR). Correctly determining the optic disc region of interest (ROI) will produce a smaller initial image which takes much lesser time taken to process compared to the entire image. The earlier ROI localization in the ARGALI system used a grid based method. The new algorithm adds a preprocessing step before analyzing the image. This step significantly improves the performance of the ROI detection. A batch of 1564 retinal images from the Singapore Eye Research Centre was used to compare the performance of the two methods. From the results, the earlier and new algorithm detects the ROI correctly for 88% and 96% of the images respectively. The results indicate potential applicability of the method for automated and objective mass screening for early detection of glaucoma.
Article
Full-text available
This paper transmits a FORTRAN-IV coding of the fuzzy c-means (FCM) clustering program. The FCM program is applicable to a wide variety of geostatistical data analysis problems. This program generates fuzzy partitions and prototypes for any set of numerical data. These partitions are useful for corroborating known substructures or suggesting substructure in unexplored data. The clustering criterion used to aggregate subsets is a generalized least-squares objective function. Features of this program include a choice of three norms (Euclidean, Diagonal, or Mahalonobis), an adjustable weighting factor that essentially controls sensitivity to noise, acceptance of variable numbers of clusters, and outputs that include several measures of cluster validity.
Article
Retinal vessel segmentation algorithms are a fundamental component of automatic retinal disease screening systems. This work examines the blood vessel segmentation methodologies in two dimensional retinal images acquired from a fundus camera and a survey of techniques is presented. The aim of this paper is to review, analyze and categorize the retinal vessel extraction algorithms, techniques and methodologies, giving a brief description, highlighting the key points and the performance measures. We intend to give the reader a framework for the existing research; to introduce the range of retinal vessel segmentation algorithms; to discuss the current trends and future directions and summarize the open problems. The performance of algorithms is compared and analyzed on two publicly available databases (DRIVE and STARE) of retinal images using a number of measures which include accuracy, true positive rate, false positive rate, sensitivity, specificity and area under receiver operating characteristic (ROC) curve.