Conference PaperPDF Available

Neighbouring Pixel Based ELM Approach for Blood Vessel Segmentation in Retinal Images

Authors:

Abstract and Figures

For early diagnosis of retinal disease, Blood vessel segmentation of retinal images play important role. For blood vessel segmentation of retinal images, we propose a supervised neighbouring pixel based ELM approach. In NP based ELM approach, maximum vessel enhanced image is obtained by contrast enhance and edge detection operation on green channel intensity. A multidimensional feature data set is constructed by neighbouring pixels with candidate pixel and target candidate pixel class from manual segmented image. Then, with the capability of universal approximation, Extreme Learning Machine (ELM) is trained with feature dataset. Finally, each candidate pixel is classified and arranged in final segmented image by post processing. Our proposed approach is validated on publicly available retinal database DRIVE. This approach is also compared with state of arts methods and has achieved the accuracy up to 0.9546.
Content may be subject to copyright.
Neighbouring Pixel Based ELM Approach for Blood
Vessel Segmentation in Retinal Images
Gaurav Jaiswal
ICT Research Lab, Department of Computer Science
University of Lucknow, Lucknow, India 226007
Email: gauravjais88@gmail.com
S P Kannojia
ICT Research Lab, Department of Computer Science
University of Lucknow, Lucknow, India 226007
Email: spkannojia@gmail.com
Abstract— For early diagnosis of retinal disease, Blood vessel
segmentation of retinal images play important role. For blood
vessel segmentation of retinal images, we propose a supervised
neighbouring pixel based ELM approach. In NP based ELM
approach, maximum vessel enhanced image is obtained by
contrast enhance and edge detection operation on green channel
intensity. A multidimensional feature data set is constructed by
neighbouring pixels with candidate pixel and target candidate
pixel class from manual segmented image. Then, with the
capability of universal approximation, Extreme Learning
Machine (ELM) is trained with feature dataset. Finally, each
candidate pixel is classified and arranged in final segmented
image by post processing. Our proposed approach is validated on
publicly available retinal database DRIVE. This approach is also
compared with state of arts methods and has achieved the
accuracy up to 0.9546.
Keywords— neighbouring pixel; vessel segmentation; extreme
learning machine; medical imaging.
I. INTRODUCTION
In medical imaging, segmentation of blood vessel of
retinal images is very important phase to treatment of retinal
disease. It helps in diagnosing abnormalities due to diabetes,
hypertension, arteriosclerosis, cardiovascular disease and
stroke [1]. Manual segmentation of blood vessel of retinal
image is time consuming, costly and it requires good training
and skills. For automated process, complexity arises due to
low contrast and background variation of retinal images [2].
To overcome these problems several methods have been
developed. These methods are basically supervised
classification methods [3][4][5][6], unsupervised classification
methods [7][8], SOM [9], matched filtering [10][11]
[12][13][29], morphological processing [14], vessel
tracing/tracking [15], multi-scale approaches [16], vessel
profile models [17] and deformable models. In supervised
methods ANN [4][5][6], k-NN [3][18], SVM [19][20][21] are
extensively investigated with PCA, Gabor filter, wavelets,
image ridges, line operator feature extraction methods. Fuzzy
C-means clustering [7][8], SOM [9], Local entropy [22], Co-
occurrence matrix methods are widely used in unsupervised
classification methods for vessel segmentation. In matched
filtering methods 2D Gaussian filtering [10][11][12][13],
threshold probing [23][24], first order derivative [13] are
extensively investigated.
Our approach is based on supervised binary classification
of pixels using neighbouring pixels feature. Here, Extreme
Learning Machine [25] is exploited as binary classifier for
pixel classification into two classes viz. vessel pixel or non-
vessel pixel. With the capability of universal approximation,
ELM is trained with feature dataset. Ten cross fold validation
method is applied for partitioning training and test data set.
Finally, each candidate pixel is classified and arranged in final
segmented image by post processing.
The remaining of paper is organized as follow. Section II
presents proposed Neighbouring pixel based ELM approach
for blood vessel segmentation in retinal images with brief
overview of neighbouring pixel and ELM. Section III presents
each step of experiments, details of DRIVE database,
comparison and results followed by the conclusion in Section
IV.
II. PROPOSED METHOD
Our approach is based on manual segmentation process of
images. Human always considers neighbourhood of region of
interest and adaptively learns. The main idea behind our
approach is reducing the image segmentation problem into
low resolution binary classification problem.
A. Neighbouring Pixel Feature Extraction
In image, neighbouring pixel [30] N8(P) is defined for any
candidate pixel P(x, y), two vertical and two horizontal
neighbours
N4(P) = (x+1, y), (x-1, y), (x, y+1), (x, y-1) (1)
Four diagonal neighbours
ND(P) = (x+1, y+1), (x+1, y-1), (x-1, y+1), (x-1, y-1) (2)
N4(P) and ND(P) jointly defines N8(P). Fig. 1 depicts
neighbouring pixel of candidate pixel (x,y).
Fig. 1. Neighbouring pixels of candidate pixel (x,y)
Then, Multidimensional feature set is constructed by
neighbourhood pixel with candidate pixel and target class,
which is obtained by corresponding pixel of candidate pixel in
manual segmented image (gold standard image).
B. Extreme Learning Machine
Extreme learning machine (ELM) was proposed for training
single hidden layer feed forward neural network (SLFNs) [26].
In ELM , the hidden nodes of hidden layer are randomly
initialised and then fixed without iteratively tuning. The only
free parameters need to be learned are connections (or
weights) between the hidden layer and output layer.
The output function of ELM for generalized SLFNs is
()=()=()
 (3)
Where β= [β1, . . . , βL]T is the output weight vector between the
hidden layer of L nodes to the m (≥1) output nodes, and
h(x)=[h1(x), . . . , hL(x)] is ELM nonlinear feature mapping.
hi(x) is the output of ith hidden node for input x [25].
Basically, ELM trains an SLFN in two main stages: random
feature mapping and linear parameters solving. In the first
stage, ELM randomly initializes the hidden layer to map the
input data into a feature space by some nonlinear mapping
functions. In the second stage of ELM learning, the weights
connecting the hidden layer and the output layer, denoted by
β, are solved by minimizing the approximation error in the
squared error sense:
∈×(4)
Where H is the hidden layer output matrix (randomized
matrix):
=()
()=()⋯ ℎ()
⋮ ⋮ ⋮
()⋯ ℎ()(5)
And T is the training data target matrix:
=
)
= ⋯ 
⋮ ⋮
 ⋯ (6)
The optimal solution to (4) is given by
=(7)
Where H denotes the MoorePenrose generalized inverse of
matrix H.
In our approach, ELM (shown in Fig. 2) is chosen as binary
classifier (m=2, d=9, L=100) due to its good generalization
performance, universal approximation capability and real time
learning feature [25].
C. Algorithm of Proposed Approach
Execution of our algorithm is divided in two phases:
1)Training Phase, 2)Testing Phase. In training phase, input
image is taken and image enhancement process is applied to
achieve the good contrast of vessel in image. Then, feature
vector is constructed for each pixel with its neighbouring
pixels and target class from manual segmented image. ELM
classifier is trained with this feature vector set of all pixels. In
testing phase, pre-processing step is same as training phase. In
feature extraction, neighbourhood pixels are extracted with
candidate pixel. The ELM classifier predicts target value for
all pixels. Post-processing step converts all predicted pixel
into image form and remove orphan pixel less than 4 pixel
connectivity to reduce noise. Fig. 3 shows all the step of
algorithm pipeline.
Fig. 2. Extreme learning machine classifier architecture
Fig. 3. Algorithm pipeline
III. EXPERIMENT AND RESULTS
A. DRIVE Database
The DRIVE database [3][27] was created by collecting
fundus photographs from a diabetic retinopathy screening
program in The Netherlands. This program was consisted of
400 diabetic subjects of age 25-90 years. 40 photographs have
been randomly chosen for DRIVE database, in which 33
photographs do not contain sign of diabetic retinopathy and 7
contains signs of early diabetic retinopathy. All 40 images are
captured using a canon CR5 non-mydriatic 3CCD camera with
a 45 degree field of view (FOV) and 32 bit RGB JPEG
compressed.
These 40 images has been divided into a training set,
which contains 20 training images, and a test set, which
contains 20 test images. For each image, a mask image is
provided that delineates the FOV. For training set, a single
manual segmentation of the blood vessel is given for each
image. For test set, two manual segmentation of the blood
vessel is given for each image.
B. Experiment
Experiment is implemented in MATLAB and performed
on KRISHNA cluster at CFCR, University of Lucknow.
Training images of DRIVE database is used to construct the
feature dataset. Green channel of image have good visibility of
blood vessel comparing to red or blue channel of image. After
selecting the green channel of image, logarithmic filter is
applied to achieve edges of vessels. To remove noise and
enhance the contrast of image, contrast adjusts and smoothing
filters are applied. Resultant image masked with mask image
for FOV.
Masked image is padded with zero on one pixel outer side
so that neighbouring pixel can be extracted for each pixel.
Intensity of neighbouring pixel with candidate pixel are
computed and stored in feature dataset with correspond
candidate pixel in manual segmented Gold standard image.
Feature dataset is used for training and validation for ELM
binary classifier. Accuracy of training and validation of Binary
classifier is increased with no. of hidden layer nodes. In this
experiment, ELM classifier is configured with 100 node of
hidden layer.
Fig. 4 shows the graph between accuracy of training and
validation of ELM binary classifier with number of hidden
layer nodes.
For segmentation of blood vessel, image is pre-processed
and extracted feature vector as above described method. The
predicted pixel value is obtained from ELM classifier that
classifies the candidate pixel in vessel pixel or non-vessel
pixel. This process is performed for each pixel of image. After
post-processing of predicted pixel, Final segmented image are
obtained.
Fig. 5 shows experimental result of proposed approach for
an image randomly chosen from DRIVE database.
C. Performance Analysis
Performance analysis of proposed approach involves
comparison of predicted image and its manual segmented
image (gold standard image).
The contingency of classification defines as:
TP: Vessel pixel detected Vessel pixel present
FP: Vessel pixel detected Vessel pixel absent
FN: Vessel pixel not detected Vessel pixel present
TN: Vessel pixel not detected Vessel pixel absent
1). Performance Metrics
Our approach is examined on these performance
evaluation metrics [31] viz. sensitivity (Se), specificity (Sp),
positive predictive value (Ppv), negative predictive value (Npv)
and accuracy (Acc).
=
+(8)
=
+(9)
 =
+(10)
 =
+(11)
Fig. 5. (a) Original RGB image of retina (b) Green channel
image of original image (c) Enhanced contrast image (d) Masked
maximum enhanced image (e) Segmented image by proposed
approach (f) Gold standard segmented Image
Fig. 4. Graph between accuracy of training and validation of
ELM binary classifier with number of hidden layer nodes
 =+
+++(12)
Table I shows performance metrics of proposed approach
for all test images of DRIVE Database.
TABLE I. PERFORMANCE METRICS OF PROPOSED APPROACH FOR ALL TEST
IMAGES OF DRIVE DATABASE
Image Se Sp Ppv Npv Acc
1 0.8156 0.9675 0.6763 0.9871 0.9573
2 0.8714 0.9665 0.7252 0.9894 0.9590
3 0.8835 0.9567 0.6100 0.9914 0.9508
4 0.9142 0.9627 0.6750 0.9949 0.9593
5 0.8949 0.9619 0.6490 0.9931 0.9575
6 0.9589 0.9430 0.4584 0.9981 0.9429
7 0.8438 0.9617 0.6376 0.9904 0.9549
8 0.8391 0.9509 0.4757 0.9927 0.9457
9 0.9117 0.9501 0.4331 0.9973 0.9490
10 0.8050 0.9711 0.6987 0.9865 0.9599
11 0.8410 0.9660 0.6812 0.9879 0.9568
12 0.8683 0.9599 0.5877 0.9924 0.9544
13 0.8973 0.9559 0.6008 0.9934 0.9520
14 0.8252 0.9683 0.6679 0.9888 0.9592
15 0.6440 0.9834 0.8406 0.9695 0.9553
16 0.8722 0.9583 0.5801 0.9928 0.9536
17 0.9120 0.9448 0.3917 0.9971 0.9435
18 0.8260 0.9623 0.5904 0.9915 0.9556
19 0.7967 0.9801 0.8050 0.9828 0.9652
20 0.8198 0.9689 0.6250 0.9902 0.9609
Average 0.8520 0.9620 0.6205 0.9904 0.9546
2). Comparison with State of Arts Methods
Experimental result is also compared with some state of
arts methods on basis of average accuracy. Table II shows
comparison of performance of our approach with state of arts
methods on Drive database. Our approach has achieved
segmentation accuracy up to 0.9546.
TABLE II. COMPARISON OF PERFORMANCE OF OUR APPROACH WITH
STATE OF ARTS METHODS ON DRIVE DATABASE
Method Method Type Accuracy on
DRIVE database
Martinez-Perez et al.[28] Rule based 0.934
Jiang and Mojon[23] Model Based 0.891
Chaudhuri et al.[10] Matched Filter 0.877
Cinsdikici and Aydin[29] Matched Filter 0.929
Niemeijer et al. [18] Supervised 0.941
Zhang et al.[9] Clustering based 0.940
Proposed approach ELM based 0.954
IV. CONCLUSION
For early diagnosis of retinal disease, Neighbouring pixel
based ELM approach for blood vessel segmentation in retinal
image can be used.
The experimental result of proposed approach is validated
on publicly available DRIVE database with gold standard
segmented image. This approach is quantitatively evaluated on
the basis of performance metrics (Se, Sp, Ppv, Npv, Acc). It is
compared with other state of arts methods on basis of average
accuracy and it outperforms with accuracy up to 0.9546.
ACKNOWLEDGMENT
We thankful to Central facility of Computational Research,
University of Lucknow for providing access to KRISHNA
cluster. First author is also thankful to UGC for providing
UGC-JRF fellowship to sustain his research.
REFERENCES
[1] J. J. Kanski and B. Bowling, Clinical ophthalmology: a systematic
approach. Elsevier Health Sciences, 2011.
[2] S. Roychowdhury, D. D. Koozekanani, and K. K. Parhi, “Blood vessel
segmentation of fundus images by major vessel extraction and subimage
classification,” IEEE journal of biomedical and health informatics, vol.
19, no. 3, pp. 1118–1128, 2015.
[3] J. Staal, M. D. Abramoff,` M. Niemeijer, M. A. Viergever, and B. van
Ginneken, “Ridge-based vessel segmentation in color images of the
retina,” IEEE transactions on medical imaging, vol. 23, no. 4, pp. 501
509, 2004.
[4] K. Akita and H. Kuga, “A computer method of understanding ocular
fundus images,” Pattern recognition, vol. 15, no. 6, pp. 431–443, 1982.
[5] R. Nekovei and Y. Sun, “Back-propagation network and its
configuration for blood vessel detection in angiograms,” IEEE
Transactions on Neural Networks, vol. 6, no. 1, pp. 64–72, 1995.
[6] C. Sinthanayothin, J. F. Boyce, H. L. Cook, and T. H. Williamson,
“Automated localisation of the optic disc, fovea, and retinal blood
vessels from digital colour fundus images,” British Journal of
Ophthalmology, vol. 83, no. 8, pp. 902–910, 1999.
[7] Y. A. Tolias and S. M. Panas, “A fuzzy vessel tracking algorithm for
retinal images based on fuzzy clustering,” IEEE Transactions on
Medical Imaging, vol. 17, no. 2, pp. 263–273, 1998.
[8] G. B. Kande, P. V. Subbaiah, and T. S. Savithri, “Unsupervised fuzzy
based vessel segmentation in pathological digital fundus images,” Jour-
nal of medical systems, vol. 34, no. 5, pp. 849–858, 2010.
[9] J. Zhang, Y. Cui, W. Jiang, and L. Wang, “Blood vessel segmentation of
retinal images based on neural network,” in International Conference on
Image and Graphics, pp. 11–17, Springer, 2015.
[10] S. Chaudhuri, S. Chatterjee, N. Katz, M. Nelson, and M. Goldbaum,
“Detection of blood vessels in retinal images using two-dimensional
matched filters,” IEEE Transactions on medical imaging, vol. 8, no. 3,
pp. 263–269, 1989.
[11] M. Al-Rawi, M. Qutaishat, and M. Arrar, “An improved matched filter
for blood vessel detection of digital retinal images,” Computers in
Biology and Medicine, vol. 37, no. 2, pp. 262–267, 2007.
[12] A. Fielder and K. Cocker, “Automatic extraction of the structure of the
retinal blood vessel network of premature infants,” J Med Assoc Thai,
vol. 90, no. 9, pp. 1780–92, 2007.
[13] B. Zhang, L. Zhang, L. Zhang, and F. Karray, “Retinal vessel extraction
by matched filter with first-order derivative of gaussian,” Computers in
biology and medicine, vol. 40, no. 4, pp. 438–445, 2010.
[14] Y. Yang, S. Huang, and N. Rao, “An automatic hybrid method for
retinal blood vessel extraction,” International Journal of Applied
Mathematics and Computer Science, vol. 18, no. 3, pp. 399–407, 2008.
[15] O. Chutatape, L. Zheng, and S. Krishnan, “Retinal blood vessel
detection and tracking by matched gaussian and kalman filters,” in
Engineering in Medicine and Biology Society, 1998. Proceedings of the
20th Annual International Conference of the IEEE, vol. 6, pp. 3144–
3149, IEEE, 1998.
[16] M. Vlachos and E. Dermatas, “Multi-scale retinal vessel segmentation
using line tracking,” Computerized Medical Imaging and Graphics, vol.
34, no. 3, pp. 213–227, 2010.
[17] V. Mahadevan, H. Narasimha-Iyer, B. Roysam, and H. L. Tanenbaum,
“Robust model-based vasculature detection in noisy biomedical
images,” IEEE Transactions on Information Technology in Biomedicine,
vol. 8, no. 3, pp. 360–376, 2004.
[18] M. Niemeijer, J. Staal, B. van Ginneken, M. Loog, and M. D. Abramoff,
“Comparative study of retinal vessel segmentation methods on a new
publicly available database,” in Medical Imaging 2004, pp. 648–656,
International Society for Optics and Photonics, 2004.
[19] E. Ricci and R. Perfetti, “Retinal blood vessel segmentation using line
operators and support vector classification,” IEEE transactions on
medical imaging, vol. 26, no. 10, pp. 1357–1365, 2007.
[20] L. Xu and S. Luo, “A novel method for blood vessel detection from
retinal images,” Biomedical engineering online, vol. 9, no. 1, p. 1, 2010.
[21] X. You, Q. Peng, Y. Yuan, Y.-m. Cheung, and J. Lei, “Segmentation of
retinal blood vessels using the radial projection and semi-supervised
approach,” Pattern Recognition, vol. 44, no. 10, pp. 2314–2324, 2011.
[22] F. M. Villalobos-Castaldi, E. M. Felipe-Riveron,´ and L. P. Sanchez´-
Fernandez,´ “A fast, efficient and automated method to extract vessels
from fundus images,” Journal of Visualization, vol. 13, no. 3, pp. 263–
270, 2010.
[23] X. Jiang and D. Mojon, “Adaptive local thresholding by verification-
based multithreshold probing with application to vessel detection in
retinal images,” IEEE Transactions on Pattern Analysis and Machine
Intelligence, vol. 25, no. 1, pp. 131–137, 2003.
[24] A. Hoover, V. Kouznetsova, and M. Goldbaum, “Locating blood vessels
in retinal images by piecewise threshold probing of a matched filter
response,” IEEE Transactions on Medical imaging, vol. 19, no. 3, pp.
203–210, 2000.
[25] G. Huang, G.-B. Huang, S. Song, and K. You, “Trends in extreme
learning machines: a review,” Neural Networks, vol. 61, pp. 32–48,
2015.
[26] G.-B. Huang, Q.-Y. Zhu, and C.-K. Siew, “Extreme learning machine:
theory and applications,” Neurocomputing, vol. 70, no. 1, pp. 489–501,
2006.
[27] M. Niemeijer, J. Staal, B. Ginneken, M. Loog, and M. Abramoff,
“Drive: digital retinal images for vessel extraction,” in Methods for
Evaluating Segmentation and Indexing techniques Dedicated to Retinal
Ophthalmology, 2004.
[28] M. E. Martinez-Perez, A. D. Hughes, S. A. Thom, A. A. Bharath, and K.
H. Parker, “Segmentation of blood vessels from red-free and fluorescein
retinal images,” Medical image analysis, vol. 11, no. 1, pp. 47–61, 2007.
[29] M. G. Cinsdikici and D. Aydın, “Detection of blood vessels in oph-
thalmoscope images using mf/ant (matched filter/ant colony) algorithm,”
Computer methods and programs in biomedicine, vol. 96, no. 2, pp. 85–
95, 2009.
[30] Dhawan, A. P., Buelloni, G., & Gordon, R. “Enhancement of
mammographic features by optimal adaptive neighborhood image
processing,” IEEE Transactions on Medical Imaging, vol. 5, no.1,pp. 8-
15, 1986.
[31] Powers, David Martin. "Evaluation: from precision, recall and F-
measure to ROC, informedness, markedness and correlation," Journal of
Machine Learning Technologies, vol. 2, no.1, pp. 37- 63, 2011.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Extreme learning machine (ELM) has gained increasing interest from various research fields recently. In this review, we aim to report the current state of the theoretical research and practical advances on this subject. We first give an overview of ELM from the theoretical perspective, including the interpolation theory, universal approximation capability, and generalization ability. Then we focus on the various improvements made to ELM which further improve its stability, sparsity and accuracy under general or specific conditions. Apart from classification and regression, ELM has recently been extended for clustering, feature selection, representational learning and many other learning tasks. These newly emerging algorithms greatly expand the applications of ELM. From implementation aspect, hardware implementation and parallel computation techniques have substantially sped up the training of ELM, making it feasible for big data processing and real-time reasoning. Due to its remarkable efficiency, simplicity, and impressive generalization performance, ELM have been applied in a variety of domains, such as biomedical engineering, computer vision, system identification, control and robotics, etc. In this review, we try to provide a comprehensive view of these advances in ELM together with its future perspectives.
Article
Full-text available
In this work we compare the performance of a number of vessel segmentation algorithms on a newly constructed retinal vessel image database. Retinal vessel segmentation is important for the detection of numerous eye diseases and plays an important role in automatic retinal disease screening systems. A large number of methods for retinal vessel segmentation have been published, yet an evaluation of these methods on a common database of screening images has not been performed. To compare the performance of retinal vessel segmentation methods we have constructed a large database of retinal images. The database contains forty images in which the vessel trees have been manually segmented. For twenty of those forty images a second independent manual segmentation is available. This allows for a comparison between the performance of automatic methods and the performance of a human observer. The database is available to the research community. Interested researchers are encouraged to upload their segmentation results to our website (http://www.isi.uu.nl/Research/Databases). The performance of five different algorithms has been compared. Four of these methods have been implemented as described in the literature. The fifth pixel classification based method was developed specifically for the segmentation of retinal vessels and is the only supervised method in this test. We define the segmentation accuracy with respect to our gold standard as the performance measure. Results show that the pixel classification method performs best, but the second observer still performs significantly better.
Article
It is clear that the learning speed of feedforward neural networks is in general far slower than required and it has been a major bottleneck in their applications for past decades. Two key reasons behind may be: (1) the slow gradient-based learning algorithms are extensively used to train neural networks, and (2) all the parameters of the networks are tuned iteratively by using such learning algorithms. Unlike these conventional implementations, this paper proposes a new learning algorithm called extreme learning machine (ELM) for single-hidden layer feedforward neural networks (SLFNs) which randomly chooses hidden nodes and analytically determines the output weights of SLFNs. In theory, this algorithm tends to provide good generalization performance at extremely fast learning speed. The experimental results based on a few artificial and real benchmark function approximation and classification problems including very large complex applications show that the new algorithm can produce good generalization performance in most cases and can learn thousands of times faster than conventional popular learning algorithms for feedforward neural networks.1
Conference Paper
Blood vessel segmentation of retinal images plays an important role in the diagnosis of eye diseases. In this paper, we propose an automatic unsupervised blood vessel segmentation method for retinal images. Firstly, a multi-dimensional feature vector is constructed with the green channel intensity and the vessel enhanced intensity feature by the morphological operation. Secondly, self-organizing map (SOM) is exploited for pixel clustering, which is an unsupervised neural network. Finally, we classify each neuron in the output layer of SOM as retinal neuron or non-vessel neuron with Otsu’s method, and get the final segmentation result. Our proposed method is validated on the publicly available DRIVE database, and compared with the state-of-the-art algorithms.
Article
This paper presents a novel three-stage blood vessel segmentation algorithm using fundus photographs. In the first stage, the green plane of a fundus image is pre-processed to extract a binary image after high-pass filtering, and another binary image from the morphologically reconstructed enhanced image for the vessel regions. Next, the regions common to both the binary images are extracted as the major vessels. In the second stage, all remaining pixels in the two binary images are classified using a Gaussian Mixture Model (GMM) classifier using a set of 8 features that are extracted based on pixel neighborhood and first and second-order gradient images. In the third post-processing stage, the major portions of the blood vessels are combined with the classified vessel pixels. The proposed algorithm is less dependent on training data, requires less segmentation time and achieves consistent vessel segmentation accuracy on normal images as well as images with pathology when compared to existing supervised segmentation methods. The proposed algorithm achieves a vessel segmentation accuracy of 95.2%, 95.15% and 95.3% in an average of 3.1 seconds, 6.7 seconds and 11.7 seconds on three public data sets DRIVE, STARE, and CHASE DB1, respectively.
Article
We have studied some fundamental problems towards the understanding of color ocular fundus images which are used in the mass diagnosis of adult diseases such as hypertension and diabetes.These problems are: the extraction of blood vessels from the retinal background; the recognition of arteries and veins; the detection and analysis of peculiar regions such as hemorrhages, exudates, optic discs and arterio-venous crossings.We propose a computer method for each of these problems and show some experimental results.
Article
Automatic segmentation of retinal blood vessels has become a necessary diagnostic procedure in ophthalmology. The blood vessels consist of two types of vessels, i.e., thin vessels and wide vessels. Therefore, a segmentation method may require two different processes to treat different vessels. However, traditional segmentation algorithms hardly draw a distinction between thin and wide vessels, but deal with them together. The major problems of these methods are as follows: (1) If more emphasis is placed on the extraction of thin vessels, the wide vessels tend to be over detected; and more artificial vessels are generated, too. (2) If more attention is paid on the wide vessels, the thin and low contrast vessels are likely to be missing. To overcome these problems, a novel scheme of extracting the retinal vessels based on the radial projection and semi-supervised method is presented in this paper. The radial projection method is used to locate the vessel centerlines which include the low-contrast and narrow vessels. Further, we modify the steerable complex wavelet to provide better capability of enhancing vessels under different scales, and construct the vector feature to represent the vessel pixel by line strength. Then, semi-supervised self-training is used for extraction of the major structures of vessels. The final segmentation is obtained by the union of the two types of vessels. Our approach is tested on two publicly available databases. Experiment results show that the method can achieve improved detection of thin vessels and decrease false detection of vessels in pathological regions compared to rival solutions.
Article
An Automatic Hybrid Method for Retinal Blood Vessel Extraction The extraction of blood vessels from retinal images is an important and challenging task in medical analysis and diagnosis. This paper presents a novel hybrid automatic approach for the extraction of retinal image vessels. The method consists in the application of mathematical morphology and a fuzzy clustering algorithm followed by a purification procedure. In mathematical morphology, the retinal image is smoothed and strengthened so that the blood vessels are enhanced and the background information is suppressed. The fuzzy clustering algorithm is then employed to the previous enhanced image for segmentation. After the fuzzy segmentation, a purification procedure is used to reduce the weak edges and noise, and the final results of the blood vessels are consequently achieved. The performance of the proposed method is compared with some existing segmentation methods and hand-labeled segmentations. The approach has been tested on a series of retinal images, and experimental results show that our technique is promising and effective.