In this project, we propose a new novel DNN-based automatic detection of diabetic retinopathy. In deep neural networks are used for classify the images that indicate diabetic retinopathy. The main aim of this project is to find the suitable way to detect the problems and classify them. We propose an deep neural network (RBFNN) classifier gives high precision in grouping of these disease through spatial examination. The RBFNN classifier does not require an large training time, therefore the model production can be expedited. We further find from our data set of 80,000 images used in our proposed RBFNN achieves a sensitivity of 95% and an accuracy of 75% on 5000 validation images. The fuzzy c means clustering is used to store the information as the processed images in this project . Finally, the proposed system is developed using matlab simulation.

1. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8937 Copyright ⓒ 2019Authors
Deep Learning Radial Basis Function Neural Networks Based
Automatic Detection of Diabetic Retinopathy
J.Friska1
; M. Mano Priya2
1
Associate Professor, Department of Electronics and communication Engineering,
2
PG Scholar, Department of Electronics and communication Engineering,
Francis Xavier Engineering College, Tirunelveli, Tamilnadu, India.
Abstract
In this project, we propose a new novel DNN-based automatic detection of diabetic
retinopathy. In deep neural networks are used for classify the images that indicate diabetic
retinopathy. The main aim of this project is to find the suitable way to detect the problems
and classify them. We propose an deep neural network (RBFNN) classifier gives high
precision in grouping of these disease through spatial examination. The RBFNN classifier
does not require an large training time, therefore the model production can be expedited. We
further find from our data set of 80,000 images used in our proposed RBFNN achieves a
sensitivity of 95% and an accuracy of 75% on 5000 validation images. The fuzzy c means
clustering is used to store the information as the processed images in this project . Finally,
the proposed system is developed using matlab simulation.
Index Terms: Deep neural networks(DNN),Radial basis function neural network (sRBFNN),
Support vector machine(SVM),Gray level co occurrence matrix(GLCM).
Introduction
Nowadays, the diagnosis has still been achieved mainly by manual techniques.
However, the accuracy of it depends on the operator’s expertise. The situation of the operator
may highly affect the analysis. Recently, there are efforts and research studies on making it
automatic the process. The works on automatization can be divided into two parts, namely,
segmentation and classification of diabetic retinopathy. The main objective of the project is to
study, implement, and classify the diabetic retinopathy problem .Since there are a lot of cells
in a diabetic retinopathy, we have to find a suitable way to detect and classify them. The
objective is to compare the sensitivity of diabetic retinopathy problem. The following
methods have some drawbacks and they are discussed.
[1] Background retinopathy occurs when diabetes damages the small blood vessels and
nerves in the retina. The retina acts like the film of the eye. It captures images coming
through the front of the eye and sends them to the brain to see. Fluids leaking from the
damaged vessels cause the retina to swell .Swelling of the macula, the central area of the
retina, can lead to vision loss. [2] In this stage, more severe and widespread changes are seen
in the retina, including bleeding into the retina. At high risk the vision could eventually be
affected. [3] In this stage, new blood vessels and scar tissue have formed on your retina,
which can cause significant bleeding and lead to retinal detachment. At high risk the vision
could be loss.[4]Retinopathy progression appears to follow different patterns. Some patients
develop leakage and other develop capillary closure which also causes loss of sight. The
number of haemorrhages and micro aneurysms indicate progression. If they increase in

2. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8938 Copyright ⓒ 2019Authors
number the retinopathy is getting worse. Dropping blood pressure to the targets will slow
down progression right away.[5]This approach is used for the task, then produces a program
that can accomplish the task of generating the correct outputs for new input . In this way, we
generate the program by a linear algorithmic process may look completely different than one
developed manually by a programmer, it may contain huge amounts of information about a
[6] A new framework for detecting retinopathy problem using support vector machine is
introduced in this paper. This paper Classify the diabetic retinopathy problem using deep
learning radial neural network classifier. Accuracy is increased by using binary searching
algorithm. Compare the sensitivity of diabetic retinopathy problem using neural network
classifier.
Proposed System
In the proposed work, we would like to propose a methodology using RBFNNs for
classifying the images that indicate diabetic retinopathy (DR). The retina images are from
fundus photography. This photography uses fundus camera to get the color images of interior
surface of the eye so that we can monitor the eye and find the disorders. Fundus camera
contains intricate microscope which attached to a flash-enabled camera. This camera helps to
photograph the interior surface of the eye including retina, macula, and posterior pole.
Currently, the DR detection is a manual and time-consuming process. The patient will go to
clinic to take fundus photograph of the retina and the image will take. He/she will get the
result in 2 or 3 days, and then patient should take the result to an ophthalmologist for the
review. As this traditional way of DR detection takes more time and the chance of
miscommunication and the delayed treatment lead is more, there we can use automatic DR
detection algorithm which can be implemented in fundus camera itself. Here, after taking the
retina image, the image will be given to already trained model which classifies the images
based on whether the patient has DR or not, if he/she has the DR, and then the severity level
of the disease. This helps both the ophthalmologist and patient so that the delayed treatment
would not be happen.
Fig 2.1 Proposed system Block diagram

3. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8939 Copyright ⓒ 2019Authors
2.1 PRE PROCESSING
The input image is preprocessed to remove noise and prepare them to be fed into
neural network. Preprocessing the input image in three steps:
 Color histogram
 RGB plane
 Filtering
2.1.1 COLOR HISTOGRAM
Here color histogram is also a histogram which even plots a graph of an image in
terms of color distribution of that particular image.
2.1.2 RGB PLANE
The input image is color image, the color processing steps converts the color image into
grayscale image. This operation is done by using the following formula.
Y = 0.299*R + 0.587*G + 0.114
2.1.3 GAUSSIAN FILTER
Filtering is a process of improving an image. Filtering can highlight some unique
features or eliminate some unwanted features. Gaussian filter conserves the edges of an
image while minimising nonspecific noises. The most commonly occurring noise is the
impulse noise. It introduces at the time of image acquisition and transmission of image is the
impulsive noise. Image noise comes from various sources. Noise can be obtained due to
communication errors and compression of images.
2.2 SEGMENTATION
Segmentation is the process of allocating some labels in to all pixels of the images
based on the similarity criteria like pixels with same label to produce meaningful areas of the
input image it is used to solve the inaccurate recognition problem. It is the major step in
object recognition in the input image. The main aim is to divide the input retina images in to
segments. This project propose Fuzzy C-means clustering based method for region based
pixel separation
2.3 FEATURE EXTRACTION
Feature extraction playing an essential job in sample characterization. Gray level
cooccurence matrix (GLCM) is used for feature extraction. The size of GLCM is decided by
the Quantities of grey dimension in the image. Extracting the features from 256 samples the
blob in the background is detected.
2.4 IMAGE CLASSIFICATION
RBFNN is utilized in the proposed work for disease classification. Neural
Networks are made out of basic components that are enlivened by organic neuron
works in parallel. We train neural system to perform explicit capacity by modifying
loads between components. The system is balanced dependent on the correlation with

4. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8940 Copyright ⓒ 2019Authors
the yield and the comparing focus until the system yield coordinates the objective.
RBFNN classifier depends on two stages, for preparing and testing. Classification
exactness relies upon preparing.
Results and Discussions
This project is implemented using Matlab simulation.
3.1 INPUT IMAGE
The figure shows input image for the proposed work. The colour conversion process is used
to convert colour image to gray scale image.
Fig 3.1 Input image for the proposed system
3.2 GRAY SCALE IMAGE
The figure shows input RGB image to gray scale image for the proposed work. This
image is given for filtering using Gaussian filter.
Fig 3.2 Gray scale converted image for proposed system
3.3 FILTERED IMAGE
The figure shows the filtered image.

5. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8941 Copyright ⓒ 2019Authors
Fig 3.3 Filtered image for the proposed system
3.4 SEGMENTED IMAGE
The figure shows the segmented image for the proposed work.The problem is recognised in
this step.
Fig 3.4 Segmented image for proposed system
3.5 BLOB BACKGROUND DETECTED
The figure shows the blob background detection
.
Fig 3.5 Blob background detection
3.6 BLOB DETECTED
The figure shows the accurate blob detection.

6. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8942 Copyright ⓒ 2019Authors
Fig 5.6 Blob detected image
3.7 DETECTED RESULT
The figure shows the great outcomes in training time, classification and mean square error.
Fig 3.7 Diabetic retinopathy problem identified image
Conclusion
This project proposed algorithm for automated feature extraction and detection
of Diabetic retinopathy provides the robust solution. Results obtained indicate that
deep learning can provide a low-cost solution for diagnosing Diabetic retinopathy
with consistency. This experiment on large dataset indicates the potential of deep
learning based model in diagnosing Diabetic retinopathy accurately from fundus
images.
Such automated system reduces dependencies on clinicians. Compared to other
algorithms, the experimental results prove that the proposed approach can detect and

7. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8943 Copyright ⓒ 2019Authors
classify the disease accurately. RBFNN implementation provides high resolution.
References
1. Yinsheng Zhang, LI Wang, Zhenguan WV, Jian Zeng, “Development of an automated screening system for
retinopathy of prematurity using a deep neural network for wide –angled retinal images”, 2019 IEEE
International Workshop.
2. P. Venkateswari, E. Jebitha Steffy, Dr. N. Muthukumaran, 'License Plate cognizance by Ocular Character
Perception', International Research Journal of Engineering and Technology, Vol. 5, No. 2, pp. 536-542,
February 2018.
3. N. Muthukumaran, Mrs R.Sonya, Dr. Rajashekhara and V. Chitra, 'Computation of Optimum ATC Using
Generator Participation Factor in Deregulated System', International Journal of Advanced Research Trends in
Engineering and Technology, Vol. 4, No. 1, pp. 8-11, January 2017.
4. B. Renuka, B. Sivaranjani, A. Maha Lakshmi, Dr. N. Muthukumaran, 'Automatic Enemy Detecting Defense
Robot by using Face Detection Technique', Asian Journal of Applied Science and Technology, Vol. 2, No. 2, pp.
495-501, April 2018.
5. Anil Lamba, "Uses Of Cluster Computing Techniques To Perform Big Data Analytics For Smart Grid
Automation System", International Journal for Technological Research in Engineering, Volume 1 Issue 7,
pp.5804-5808, 2014.
6. Anil Lamba, “Uses Of Different Cyber Security Service To Prevent Attack On Smart Home Infrastructure",
International Journal for Technological Research in Engineering, Volume 1, Issue 11, pp.5809-5813, 2014.
7. Anil Lamba, "A Role Of Data Mining Analysis To Identify Suspicious Activity Alert System”, International
Journal for Technological Research in Engineering, Volume 2 Issue 3, pp.5814-5825, 2014.
8. Anil Lamba, "To Classify Cyber-Security Threats In Automotive Doming Using Different Assessment
Methodologies”, International Journal for Technological Research in Engineering, Volume 3, Issue 3, pp.5831-
5836, 2015.
9. Anil Lamba, “A Study Paper On Security Related Issue Before Adopting Cloud Computing Service Model”,
International Journal for Technological Research in Engineering, Volume 3, Issue 4, pp.5837-5840, 2015.
10. N. Muthukumaran, 'Analyzing Throughput of MANET with Reduced Packet Loss', Wireless Personal
Communications, Vol. 97, No. 1, pp. 565-578, November 2017.
11. R. Sudhashree, N. Muthukumaran, 'Analysis of Low Complexity Memory Footprint Reduction for Delay
and Area Efficient Realization of 2D FIR Filters', International Journal of Applied Engineering Research, Vol.
10, No. 20, pp. 16101-16105, 2015.
12. F.M.Aiysha Farzana, Hameedhul Arshadh. A, Ganesan. J, Dr. N. Muthukumaran, 'High Performance VLSI
Architecture for Advanced QPSK Modems', Asian Journal of Applied Science and Technology, Vol. 3, No. 1, pp.
45-49, January 2019.
13. A.Srinithi, E.Sumathi, K.Sushmithawathi, M.Vaishnavi, Dr. N. Muthukumaran, 'An Embedded Based
Integrated Flood Forecasting through HAM Communication', Asian Journal of Applied Science and Technology,
Vol. 3, No. 1, pp. 63-67, January 2019.
14. N. Muthukumaran and R. Ravi, 'Hardware Implementation of Architecture Techniques for Fast Efficient
loss less Image Compression System', Wireless Personal Communications, Volume. 90, No. 3, pp. 1291-1315,
October 2016.
15. F.M. Aiysha Farzana, Hameedhul Arshadh. A, Sara Safreen. M, Dr. N. Muthukumaran, 'Design and
Analysis for Removing Salt and Pepper Noise in Image Processing', Indo-Iranian Journal of Scientific Research,
Vol. 3, No. 1, pp. 42-47, January 2019.
16. J. Keziah, N. Muthukumaran, 'Design of K Band Transmitting Antenna for Harbor Surveillance Radar
Application', International Journal on Applications in Electrical and Electronics Engineering, Vol. 2, No. 5, pp.
16-20, May 2016.
17. Dr. N. Muthukumaran, Dr. R. Joshua Samuel Raj, Arumugathammal. E, Karthika. N, Karthika. S,
Sangeetha. M, 'Design of Underground Mine Detecting Robot using Sensor Network', International Journal of
Emerging Technology and Innovative Engineering, Volume 5, Issue 7, pp. 519-524, July 2019.

8. Our Heritage
ISSN: 0474-9030
Vol-68-Issue-1-January-2020
P a g e | 8944 Copyright ⓒ 2019Authors
18. VP. Anubala, N. Muthukumaran and R. Nikitha, „Performance Analysis of Hookworm Detection using
Deep Convolutional Neural Network‟, 2018 International Conference on Smart Systems and Inventive
Technology (ICSSIT), pp. 348-354, 2018, doi: 10.1109/ICSSIT.2018.8748645.
19. Anil Lamba, "Uses Of Artificial Intelligent Techniques To Build Accurate Models For Intrusion Detection
System”, International Journal for Technological Research in Engineering, Volume 2, Issue 12, pp.5826-5830,
2015.
20. Anil Lamba, "Mitigating Zero-Day Attacks In IOT Using A Strategic Framework", International Journal
for Technological Research in Engineering, Volume 4, Issue 1, pp.5711-5714, 2016.
21. Anil Lamba, "Identifying & Mitigating Cyber Security Threats In Vehicular Technologies", International
Journal for Technological Research in Engineering, Volume 3, Issue 7, pp.5703-5706, 2016.
22. Anil Lamba, "S4: A Novel & Secure Method For Enforcing Privacy In Cloud Data Warehouses",
International Journal for Technological Research in Engineering, Volume 3, Issue 8, pp.5707-5710, 2016.
23. Anil Lamba, “Analysing Sanitization Technique of Reverse Proxy Framework for Enhancing Database-
Security”, International Journal of Information and Computing Science, Volume 1, Issue 1, pp.30-44, 2014.
24. N. Muthukumaran and R. Ravi, 'The Performance Analysis of Fast Efficient Lossless Satellite Image
Compression and Decompression for Wavelet Based Algorithm', Wireless Personal Communications, Volume.
81, No. 2, pp. 839-859, March 2015.
25. Jayaraman.G, Dr. N. Muthukumaran, Vanaja.A, Santhamariammal.R, 'Design and Analysis the Fire
Fighting Robot', International Journal of Emerging Technology and Innovative Engineering, Volume 5, Issue 9,
pp. 690-695, September 2019.
26. R. Joshua Samuel Raj, T.Sudarson Rama Perumal, N.Muthukumaran, „Road Accident Data Analytics
Using Map – Reduce Concept‟, International Journal of Innovative Technology and Exploring Engineering,
Volume-8, Issue-11, pp. 1032- 1037, September 2019.
27. Boselin Prabhu S. R. and Balakumar N., “Enhanced Clustering Methodology for Lifetime Maximization in
Dense WSN Fields”, International Journal for Technological Research in Engineering, Volume 4, Issue 2,
pp.343-348, October-2016
28. Boselin Prabhu S. R. and Balakumar N., “Suggested Mechanisms for the Employment of MPPT Principle
Over a Photovoltaic Module”, International Journal of Research in Electrical Engineering, Volume 3, Issue 3,
pp. 45-49, October 2016.
29. Boselin Prabhu S. R. and Balakumar N., “A Research on Various Maximum Power Point Tracking
Algorithms in a Photovoltaic System”, South Asian Journal of Engineering and Technology, Volume 2, Number
28, 1-8.
30. Boselin Prabhu S. R. and Balakumar N., “Highly Distributed and Energy Efficient Clustering Algorithm
for Wireless Sensor Networks”, International Journal of Research –Granthaalayah, Volume 4, Number 9,
September 2016.
31. Boselin Prabhu S. R. and Balakumar N., “Evaluation of Quality in Network and Interoperable Connectivity
between IP Networks”, International Journal of Current Engineering and Scientific Research, Volume 3, Issue
9, pp. 81-85.
32. Boselin Prabhu S. R. and Balakumar N., “Enhanced Zone-Based Clustering Method for Energy Efficient
Wireless Sensor Network”, ARC International Journal of Innovative Research in Electronics and
Communications, Volume 3, Issue 4, pp. 01-06, 2016.
33. Boselin Prabhu S. R. and Balakumar N., “Real-World Wireless Power Transmission under Various
Scenarios and Considerations”, International Journal of Innovative and Applied Research, Volume 4, Issue 7,
pp. 24-29.
34. Banumathi.A, Banupriya.A, Niranjana.R, Jayaraman.G, Dr. N. Muthukumaran, 'Advanced Illumination
Measurement System in Highways', Asian Journal of Applied Science and Technology, Vol. 3, No. 1, pp. 39-44,
January 2019.
35. Mrs. S. Murine Sharmili, Dr. N. Muthukumaran, 'Performance Analysis of Elevation & Building Contours
Image using K-Mean Clustering with Mathematical Morphology and SVM', Asian Journal of Applied Science
and Technology, Vol. 2, No. 2, pp. 80-85, April 2018.
36. R. Gargeya and T. Leng.-“Automated identification of diabetic retinopathy using deep learning.
Ophthalmology”, 124(7):962–969,2017.