IRJET- Detection of Cataract by Statistical Features and Classification

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 03 | Mar-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 2398
DETECTION OF CATARACT BY STATISTICAL FEATURES AND
CLASSIFICATION
E. Deepika1, Dr. S. Maheswari2, Mr. S. K. Logesh3
1PG Scholar, Department of EEE, Kongu Engineering College, Erode-638 060
2Professor (Senior Grade),Department of EEE, Kongu Engineering College,Erode-638060
3Assistant Professor, of EEE, Kongu Engineering College, Erode-638 060
------------------------------------------------------------------------***----------------------------------------------------------------------
Abstract- Cataract is the major cause of blindness in the
world and the most prevalent ocular disease. Increased risk of
cataract development is associated with UV exposure, steroid
use, diabetes, and smoking. This process cannot be reversed,
but a healthy lifestyle may slow the progression. Earlier
diagnosis of cataract will prevent vision loss. In this paper a
new method has been proposed to diagnosis of cataract using
statistical features and its severity has been classified using K-
means and ANFIS classifier.
Keywords: Fundus image, AHE, Thresholding, K-means,
ANFIS
1. Introduction
Image processing and analysis is a vast area of application,
one such field is the medicine. From olden days ,humans get
affected by the variety of diseases especially in the case of
delicate organs like eyes. One such eye disease is called
cataract which has no initial symptoms leads to the
blindness. Cataract means clouding of lens, starts to
decreasing the vision which prevents entering the light into
the retina shown in figure 1.1. It is the first and foremost
diseases causing blindness. It is different from the color
blindness. The main causes are sugar, blood pressure, aging
factors, consumption of alcohol, exposure to UV radiation
etc., There are 3 types namely nuclear, cortical and
posterior. The most common type is nuclear cataract which
forms in the centre of lens and the nucleus become pale or
yellow color. When eye get injured by some infection also
the occurrence of cataract is possible called as Traumatic
cataract. A cortical cataract is another type occurs in cortex
of lens , which surrounds the central nucleus.
Figure 1.1 Cataract Diseases
Neha Naik, Namrata Deshmukh[1] Cataract is a clouding
of the lens of the eye and occurs frequently in older age
groups. In proposed system, diagnosis will be obtained
using image processing and mining techniques on fundus
image. Feature extraction using DCT. K-NN classification
algorithm will be used to classify the image in a specific
class.
Sreejaya, Merlin K Mathew[2] Cataract can affect single or
both eyes. The symptoms may include cloudy or blurred
vision, colors that may not appear as bright as they once did,
glare, poor night vision. A nuclear cataract forms deep in the
central zone (nucleus) of the lens in eyes which is the most
common type of cataract. This paper gathered and analyzed
various techniques and concepts for cataract detection.
Yew Chung Chow, Xinting Gao[3] Cataract diagnosis in
human grading is subjective and time-consuming .
Automatic detection based on texture and intensity analysis
is proposed to address the problems of existing methods
and improve the performance from three aspects, namely
ROI detection, lens mask generation and opacity detection.
In the detection method, image clipping and texture analysis
are applied to overcome the over-detection problem for
clear lens images.
Jagadish Nayak[4] Cataract is an eye disorder, will lead to
reduced eyesight. If cataract is not treated in proper time,
then it will lead to blindness. The features of the optical eye
image such as Big Ring Area (BRA), Small Ring Area (SRA),
Edge Pixel Count (EPC) and Object Perimeter are extracted
in proposed method. These features are then used in
automatic classifier namely SVM.
2. Proposed Method
The entire process to detect the earlier diagnosis of
cataract which are arranged as follows: pre processing
method namely Adaptive Histogram Equalization is used in
diagnosis process. Thresholding technique is used for
feature extraction. The obtained features are given for
automatic classifier namely K-means and ANFIS and
classified as normal, mild and severe. The block diagram of
the proposed scheme is shown in fig1.2.

Figure 2.1 Block Diagram of Proposed Work
A. DIAGONOSIS OF CATARACT
The statistical features like mean, entropy, standard
Deviation, area, uniformity has been calculated by
thresholding technique for cataract diagnosis.
 Pre-processing
Preprocessing is performed to obtain noise free and
enhanced image, which can be used to detect features.
Preprocessing of poor quality images can ensure adequate
level of success in the automated abnormality detection. The
color images are converted to the gray scale images, because
the retinal abnormalities have better visualization in the
gray scale when compared to others. Then the
preprocessing techniques are applied to the gray scale
images which performs image enhancement.
Here, Adaptive Histogram Equalization(AHE) has been
proposed. It is a contrast enhancement technique to avoid
amplifying the noise which is present in image. AHE is
different from histogram equalization .The pre-processed
output has been given for feature extraction.
 Feature Extraction
The preprocessed output has been given as input for feature
extraction . Here, Thresholding technique is used for
statistical feature extraction. The statistical features like
mean, entropy, standard deviation, intensity, area,
uniformity has been obtained. The threshold value for mean,
entropy ,uniformity ,standard deviation, area and intensity.
For all these features, above the threshold value there will
be a detected as possibility and occurrence of cataract.
3. Classification
Classification involves the data which is used to assign the
corresponding levels with respect to groups with
homogenous characteristics with the aim of discriminating
multiple objects from each other within the image
.
 K-means clustering
K-means is the simplest unsupervised learning algorithm
which solves the clustering problem. It is a vector
quantization method which aims to partition ‘n’ observation
into K-clusters. The flowchart of K-means is shown in figure
3.1. It produce an competitive results and hence used in
image segmentation.
Figure 3.1 Flow chart of K-means clustering
 ANFIS
Adaptive Neural Fuzzy Interface System is an adaptive
network. An adaptive network is a network of nodes and
directional links. This network is called adaptive because
some or all of the nodes have parameters which affect the
output of node. The ANFIS architecture is shown in figure
3.2. The circular nodes represent nodes that are fixed
whereas the square nodes are nodes that have parameters
to be learnt.

Figure 3.2 ANFIS architecture
For the training of a network, there is a forward pass and a
backward pass. The forward pass propogates the input
vector through the network layer by layer. In the backward
pass, the error is sent back through the network in a similar
manner to back propagation.
4. Experimental Results
A. Results of Pre processing
The experimental results of pre processing by applying AHE
as shown in figure 4.1 and 4.2
(a) Original image (b) AHE image
Figure 4.1 and 4.2 original and pre processed output
B. Statistical Features
The statistical features are obtained for normal and affected
images by applying thresholding from the pre processed
output as tabulated in table 4.1 and 4.2 as shown below. The
standard deviation is similar for both healthy and affected
images, so the values of SD obtained need not be considered
for classification process.
Table 4.1 Statistical Features for Healthy Images
Table 4.2 Statistical features for Affected Images
C. Classification results
1) K-means Clustering
The obtained statistical feature is given as input for the
ANFIS and K-means algorithm. The obtained output for K-
means algorithm is shown in figure 4.3. The color violet -
normal, yellow- mild and blue-Severe.
Figure 4.3 Centroid calculation and clustering output

Table 4.3 Number of training and testing data for K-means
From the obtained result the sensitivity and specificity are
calculated by the formula,
Sensitivity= TP / (TP+FN)
Specificity= TN /(TN+FP)
Accuracy= TP+TN / TP+TN+FP+FN
Where,TP-True Positive
TN-True Negative
FP-False Positive
FN-False Negative
The obtained sensitivity and specificity are Accuracy
=98.57%, Sensitivity = 92% and Specificity =95.33%.
2) ANFIS Classification
The ANFIS classification output is shown in figure 4.4.The
output is classified as normal, mild and severe stages.
Figure 4.4 ANFIS Classification output
The number of training and testing data are tabulated in
table 4.4
Table 4.4 Number of training and testing data for ANFIS
From the table the obtained sensitivity and specificity are
shown below: Accuracy=97.77% Sensitivity=85.7%
Specificity=93%.
5. Conclusion
This work developed a earlier diagnosis for cataract
detection. The fundus image was enhanced by Adaptive
Histogram Equalization. The statistical features as mean,
entropy, standard deviation, area, uniformity, intensity
were extracted by thresholding technique. The extracted
features were given as input for K-means and ANFIS
classification. From the experimental results of the proposed
system, accuracy, sensitivity and sensitivity for K-means
clustering produced good results. The average accuracy of
the tested results for the K-means and ANFIS are 98.57 and
97.77 respectively. Also, the average sensitivity, specificity,
for the K-means and ANFIS classifier tested images are
92,95.33,85.7 and 93 respectively The obtained sensitivity,
specificity and accuracy were better for K-means clustering
algorithm. In future various algorithms has been used to
detect the earlier cataract detection
References
[1] Neha Naik, Namrata Deshmukh” Eye Disease Detection
Using Computer Vision” in International Journal on Recent
and Innovation Trends in Computing and Communication on
December 2016
[2] Sreejaya, Merlin K Mathew “ Various Cataract Detection
Methods-A Survey” in International Research Journal of
Engineering and Technology on Jan 2017
[3] Visual Impairment and Blindness. Available online:
http://www.who.int/mediacentre/factsheets/fs282/ en/
(accessed on 20 April 2016).
[4] Gary, B.; Taylor, H. Cataract Blindness–Challenges for
21st century. Bull. World Health Organ. Available online:
http://www.who.int/bulletin/archives/79(3)249.pdf
(accessed on 30 April 2016).
[5] Comas, O.; Cotin, S.; Duriez, C. A shell model for real time
simulation of intra-ocular implant deployment. In
Proceedings of the International Symposium on Biomedical
Simulation, Phoenix, AZ, USA, 23–24 January 2010; pp. 160–
170.
[6] Supriyanti, R.; Habe, H.; Kidode, M.; Nagata, S. A simple
and robust method to screen cataract using specular
reflection appearance. In Proceedings of the Medical
Imaging International Conference of International Society
for Optics and Photonics (SPIE), San Diego, CA, USA, 17
March 2008. [CrossRef]
STAGES No. of training
data
No. of
testing data
Normal 20 19
Mild 15 10
Severe 15 12
STAGES No. of training
data
No. of
testing data
Normal 20 15
Mild 15 10
Severe 15 12

[7] Supriyanti, R.; Habe, H.; Kidode, M.; Nagata, S. Cataract
Screening by Specular Reflection and Texture Analysis. In
Proceedings of the Systemics and Informatics World
Network (SIWN 2009), Leipzig, Germany, 23–25 March
2009.
[8] Shashwat Pathak * and Basant Kumar “ A Robust
Automated Cataract Detection Algorithm Using Diagnostic
Opinion Based Parameter Thresholding for Telemedicine
Application”
[9] H. Li, L. Ko, J. H. Lim, J. Liu, D. W. K. Wong, T. Y. Wong and
Y. Sun, “Image Based Diagnosis of Cortical Cataract,”
Proceedings of Annual International Conference of the IEEE
Engineering in Medicine and Biology Society, 2008.
[10] Pooja Sachdeva and KiranJot Singh, “Automatic
Segmentation and Area Calculation of Optic Disc in
Ophthalmic Images”,2015 2nd International Conference on
Recent Advances in Engineering & Computational Sciences
(RAECS)
[11] Yunendah Nur Fuadah, Agung W. Setiawan, Tati L.R.
Mengko, and Budiman, “A computer aided healthcare system
for cataract classification and grading based on fundus
image analysis”, Elsevier Science Publishers B. V.
Amsterdam, The Netherlands, The Netherlands ,May 2015
[12] Yunendah Nur Fuadah, Agung W. Setiawan, Tati L.R.
Mengko, ” Mobile Cataract Detection using Optimal
Combination of Statistical Texture Analysis”,2015 4th
International Conference on Instrumentation,
Communications, Information Technology, and Biomedical
Engineering (ICICI-BME)
[13] Li, Huiqi, Joo Hwee Lim, Jiang Liu, Paul Mitchell, Ava
Grace Tan, Jie Jin Wang, and Tien Yin Wong. 2010. A
computer-aided diagnosis system of nuclear cataract. IEEE
Transactions on Biomedical Engineering
[14] Chow, Yew Chung, Xinting Gao, Huiqi Li, Joo Hwee Lim,
Ying Sun, and Tien Yin Wong.” Automatic detection of
cortical and PSC cataracts using texture and intensity
analysis on retroillumination lens images” IEEE Engineering
in Medicine and Biology Society 2011
[15]https://www5.cs.fau.de/research/data/fundus-images

IRJET- Detection of Cataract by Statistical Features and Classification

More Related Content

What's hot (19)

Similar to IRJET- Detection of Cataract by Statistical Features and Classification (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Detection of Cataract by Statistical Features and Classification