RETINAL IMAGE CLASSIFICATION USING NEURAL NETWORK BASED ON A CNN METHODS

RETINAL IMAGE CLASSIFICATION USING NEURAL NETWORK BASED ON A
CNN METHODS
G. MadhuMithra, G. Sasikala, A.Tharamaraiselvi
PG Scholar, Assistant Professor, Assistant Professor
Department of Computer Science and Engineering
Vivekananda College of Engineering for Women
-----------------------------------------------------------------------------***----------------------------------------------------------------------
ABSTRACT: This article is to develop a system for
distinguishing retinal diseases from fundus images.
Accurate and programmed analysis of retinal images is
believed to be an effective method for determining retinal
disorders such as diabetic retinopathy, hypertension and
atherosclerosis. In this task, a convoluted neural network is
used to extract various retinal features such as the retina,
optic nerve, and lesions, and to detect multiple retinal
diseases in fundus photographs participating in a
structured analysis (STARE) database of the retina. I
applied the base model. He described an innovative
solution that used convolutional neural networks (CNNs) to
enable efficient disease detection and deep learning, with
great success in the classification of various retinal
diseases. Various neural and slice-by-slice visualization
techniques were applied using CNNs trained on publicly
available retinal disease image datasets. Neural networks
have been observed to be able to capture the color and
texture of disease-specific lesions at diagnosis. This is
similar to human decision making. And this model for
deploying the Django web framework. Experiment with
various retinal features as input to a convolutional neural
network for effective classification of retinal images.
Keywords: Retinal, deep learning, TensorFlow, Keras,
CNN
1. INTRODUCTION
Data science uses scientific methods, processes,
algorithms, and systems to extract knowledge and insights
from structured and unstructured data, and apply
knowledge and practical insights from data in a variety of
application disciplines. It is an interdisciplinary field to do.
The term "data science" can be traced back, when Peter
Naur proposed it as an alternative to computer science. In
1996, the International Association of Classification
Societies was the first conference dedicated to data science.
But the definition was still in flux.
The term "data science" was first coined by DJ. Patil
and Jeff Hammerbacher, pioneers in data and analytics
efforts on LinkedIn and Facebook. In less than a decade, it
has become one of the hottest and trendiest professions in
the world. Data science is a research discipline that combines
subjectual knowledge, programming skills, and knowledge of
mathematics and statistics to derive meaningful insights
from data. Data science can be defined as a combination of
mathematics, business sense, tools, algorithms and machine
learning techniques. All of this helps reveal insights and
patterns hidden from raw data that are very helpful in
initiating good business decisions.
Data scientists find out which questions need to be
answered and where the relevant data is. In addition to
business insight and analytical skills, you have the ability to
mine, clean up, and present data. Enterprises use data
scientists to procure, manage, and analyze large amounts of
unstructured data.
Required Skills for a Data Scientist
Programming: Python, SQL, Scala, Java, R,
Machine Learning: Natural Language Processing,
Classification, Clustering. Data Visualization: Tableau, SAS,
D3.js, Python, Java, R libraries. Big data platforms: MongoDB,
Oracle, Microsoft Azure, Cloudera.
Model Preprocessing and Training (CNN): The dataset
is preprocessed. B. Image conversion, resizing, and
conversion to array format. The same process is performed
on the test image. For example, a dataset consisting of images
of retinal disease. Each image can be used as a software test
image.
Raw
image
Build a
sequentia
l model
CNN train
CNN
Weights
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 409

The training dataset is used to train the model (CNN).
This allows the model to identify the test image and the
disease. CNN has various layers such as Dense, Dropout,
Activation, Flatten, Convolution2D and MaxPooling2D. After
the model has been successfully trained, the software can
identify the disease if a retinal image is included in the
dataset. After successful training and pretreatment, the test
image and the trained model are compared to predict the
disease.
Requirements are the basic constraints needed to
develop a system. Requirements are collected during
system design. The following are the requirements to be
discussed.
1. Functional requirements
2. Non-Functional requirements
3. Environment requirements
Functional requirements:
Software requirements specifications are technical
specifications of software product requirements. This is the
first step in the requirements analysis process. Lists the
requirements for a particular software system. The
following details follow special libraries such as
TensorFlow, Keras, Matplotlib.
Non-Functional Requirements:
Process of functional steps,
1. Problem define
2. Preparing data
3. Evaluating algorithm
4. Improving results
BACKGROUND STUDY
Carlos Hernandez-Matasa , Antonis A. Argyrosa,b
, Xenophon Zabulisa [1] The first fundus images were
acquired after the invention of the ophthalmoscope. The
concept of storing and analyzing retinal images for
diagnostic purposes exists ever since. The first work on
retinal image processing was based on analog images and
regarded the detection of vessels in fundus images with
fluorescein . The fluorescent agent enhances the
appearance of vessels in the image, facilitating their
detection and measurement by the medical professional or
the computer. However, fluorescein angiography is an
invasive and time-consuming procedure and is associated
with the cost of the fluorescent agent and its
administration. Digital imaging and digital image
processing have proliferated the use of retinal image analysis
in screening and diagnosis. The ability to accurately analyze
fundus images has promoted the use of noninvasive, fundus
imaging in these domains. Moreover, the invention of new
imaging modalities, such as optical coherence tomography
(OCT) and scanning laser ophthalmoscopy (SLO), has
broadened the scope and applications of retinal image
processing. This review regards both fundus imaging, as
implemented by fundus photography and SLO and OCT
imaging.
[2] Rubina sarki , Khandakar Ahmed , (senior
member, IEEE), Hua wang , (Member, IEEE), and Yanchun
zhang Diabetes Mellitus, or Diabetes, is a disease in which a
person’s body fails to respond to insulin released by their
pancreas, or it does not produce sufficient insulin. People
suffering from diabetes are at high risk of developing various
eye diseases over time. As a result of advances in machine
learning techniques, early detection of diabetic eye disease
using an automated system brings substantial benefits over
manual detection. A variety of advanced studies relating to
the detection of diabetic eye disease have recently been
published. This article presents a systematic survey of
automated approaches to diabetic eye disease detection from
several aspects, namely: i) available datasets, ii) image
preprocessing techniques, iii) deep learning models and iv)
performance evaluation metrics. The survey provides a
comprehensive synopsis of diabetic eye disease detection
approaches, including state of the art field approaches, which
aim to provide valuable insight into research communities,
healthcare professionals and patients with diabetes.
Baidaa Al-Bander [3] The interpretation of
ophthalmic images is typically performed by trained clinical
experts. However, due to the volume and complexity of these
images, and the large variation in pathology, in addition to
the variation among experts, there has been increasing
interest in computer-assisted assessment and diagnosis of
such images. There has been particular interest in finding a
cost-effective approach with high sensitivity and specificity,
independent of human intervention, and robust enough to be
applied to large populations in a timely manner to identify
retinal diseases. This thesis introduces novel deep learning
methodologies based on convolutional neural networks
(CNNs) to address key challenges in different retinal image
analysis tasks. Three retinal image analysis objectives have
been considered in this research project: fovea and optic disc
(OD) localisation, choroid and optic disc/cup segmentation,
and disease and lesion classification tasks. In the first retinal
image analysis task, simultaneous detection of the centres of
the fovea and the optic disc from colour fundus images is
considered as a regression problem.

Suvajit Dutta, Bonthala CS Manideep, Syed Muzamil
Basha, Ronnie D. Caytiles1 and N. Ch. S. N. Iyengar2 [4]
Diabetes or more precisely Diabetes Mellitus (DM) is a
metabolic disorder happens because of high blood sugar
level in the body. Over the time, diabetes creates eye
deficiency also called as Diabetic Retinopathy (DR) causes
major loss of vision. The symptoms can originate in the
retinal area are augmented blood vessels, fluid drip,
exudates, hemorrhages, and micro aneurysms. In modern
medical science, images are the indispensable tool for
precise diagnosis of patients. In the meantime evaluation of
contemporary medical imageries remains complex. In
recent times computer vision with Deep Neural Networks
can train a model perfectly and level of accuracy also will be
higher than other neural network models. In this study
fundus images containing diabetic retinopathy has been
taken into consideration. The idea behind this paper is to
propose an automated knowledge model to identify the key
antecedents of DR. Proposed Model have been trained with
three types, back propagation NN, Deep Neural Network
(DNN) and Convolutional Neural Network (CNN) after
testing models with CPU trained Neural network gives
lowest accuracy because of one hidden layers whereas the
deep learning models are out performing NN. The Deep
Learning models are capable of quantifying the features as
blood vessels, fluid drip, exudates, hemorrhages and micro
aneurysms into different classes. Model will calculate the
weights which gives severity level of the patient’s eye. The
foremost challenge of this study is the accurate verdict of
each feature class thresholds.
Carlos Hernandez-Matasa , Antonis A. Argyrosa,b ,
Xenophon Zabulisa [5] The diabetes retinopathy is the
application of medical image processing. The retinal images
are evaluated to diagnose the DR. It is however, time
consuming and resource demanding to manually grade the
images such that the severity of DR can be defined. When
the tiny blood vessels present within the retina are
damaged, only then can one notice this problem. Blood will
flow from this tiny blood vessel and features are formed
from the fluid that exists on retina. The kinds of features
involved here due to the leakage of fluid and blood from the
blood vessels are considered to be the most important
factors to study this problem. The diabetes retinopathy
detection techniques has the three phase which pre-
processing, segmentation and classification. In this work,
NN approach is used for the classification of diabetes
portion from the image. The proposed model is
implemented in MATLAB and results are analyzed in terms
of certain parameters.
PROPOSED METHODOLOGY:
The proposed approach consists stage
pretreatments were performed in retinal images from data
sets and standardize them to size then classification was
made by Convolutional Neural Network which is a deep
learning algorithm and success was achieved to deep
learning technique so that a person with lesser expertise in
software should also be able to use it easily. It proposed
system to predicting retinal disease. It explains about the
experimental analysis of Samples of images are collected that
comprised of different retinal. The primary attributes of the
image are relied upon the shape and texture oriented
features. An efficient disease detection and deep learning
with convolutional neural networks (CNNs) has achieved
great success in the classification of various retinal diseases.
A variety of neuron-wise and layer-wise visualization
methods were applied using a CNN, trained with a publicly
available retinal disease given image dataset. The sample
screenshots displays the retinal disease detection using color
based classification model. And to deploy this model in web
application for Django framework.
Increasing throughput & reducing subjectiveness
arising from human experts in detecting the retinal decease.
It is essential to detect a particular disease. In our country
many farmers are not so educated to get correct information
about all diseases.
Here is an overview of what we are going to cover:
1. Installing the Python anaconda platform.
2. Loading the dataset.
3. Summarizing the dataset.
4. Visualizing the dataset.
5. Evaluating some algorithms
6. Making some predictions.
Download and install anaconda and get the most
useful package for machine learning in Python. Load a
dataset and understand its structure using statistical
summaries and data visualization.
Machine learning models, pick the best and build confidence
that the accuracy is reliable.
Python is a popular and powerful interpreted
language. Unlike R, Python is a complete language and
platform that you can use for both research and development
and developing production systems. There are also a lot of

modules and libraries to choose from, providing multiple
ways to do each task. overwhelming.
When you are applying machine learning to your own
datasets, you are working on a project. A machine learning
project may not be linear, but it has a number.
The best way to really come to terms with a new
platform or tool is to work through a machine learning
project end-to-end and cover the key steps. Namely, from
loading data, summarizing data, evaluating algorithms and
making some predictions.
Deploying the model in Django Framework and
predicting output
In this module the trained deep learning model is
converted into hierarchical data format file (.h5 file) which
is then deployed in our django framework for providing
better user interface and predicting the output whether the
given OCT image is CNV / DME / DRUSEN / NORMAL.
DJANGO
Django is a high-level Python web framework that
enables rapid development of secure and maintainable
websites. Built by experienced developers, Django takes
care of much of the hassle of web development, so you can
focus on writing your app without needing to reinvent the
wheel. It is free and open source, has a thriving and active
community, great documentation, and many options for free
and paid-for support. Django helps you write software that
is:
COMPLETE
Django follows the "Batteries included" philosophy
and provides almost everything developers might want to
do "out of the box". Because everything you need is part of
the one "product", it all works seamlessly together, follows
consistent design principles, and has extensive and up-to-
date documentation.
Versatile
Django can be (and has been) used to build almost
any type of website — from content management systems
and wikis, through to social networks and news sites. It can
work with any client-side framework, and can deliver
content in almost any format (including HTML, RSS feeds,
JSON, XML, etc). The site you are currently reading is built
with Django!
Internally, while it provides choices for almost any
functionality you might want (e.g. several popular
databases, templating engines, etc.), it can also be extended to
use other components if needed.
SECURE
Django helps developers avoid many common
security mistakes by providing a framework that has been
engineered to "do the right things" to protect the website
automatically. For example, Django provides a secure way to
manage user accounts and passwords, avoiding common
mistakes like putting session information in cookies where it
is vulnerable (instead cookies just contain a key, and the
actual data is stored in the database) or directly storing
passwords rather than a password hash. A password hash is a
fixed-length value created by sending the password
through a cryptographic hash function.
Django can check if an entered password is correct
by running it through the hash function and comparing the
output to the stored hash value. However due to the "one-
way" nature of the function, even if a stored hash value is
compromised it is hard for an attacker to work out the
original password. Django enables protection against many
vulnerabilities by default, including SQL injection, cross-site
scripting, cross-site request forgery and clickjacking
(see Website security for more details of such attacks).
SCALABLE
Django uses a component-based “shared-nothing”
architecture (each part of the architecture is independent of
the others, and can hence be replaced or changed if needed).
Having a clear separation between the different parts means
that it can scale for increased traffic by adding hardware at
any level: caching servers, database servers, or application
servers. Some of the busiest sites have successfully scaled
Django to meet their demands (e.g. Instagram and Disqus, to
name just two).
MAINTAINABLE
Django code is written using design principles and
patterns that encourage the creation of maintainable and
reusable code. In particular, it makes use of the Don't Repeat
Yourself (DRY) principle so there is no unnecessary
duplication, reducing the amount of code. Django also
promotes the grouping of related functionality into reusable
"applications" and, at a lower level, groups related
code into modules (along the lines of the Model View
Controller (MVC) pattern).

PORTABLE
Django is written in Python, which runs on many
platforms. That means that you are not tied to any particular
server platform, and can run your applications on many
flavours of Linux, Windows, and Mac OS X.
Furthermore, Django is well-supported by many web
hosting providers, who often provide specific infrastructure
and documentation for hosting Django sites.
RESULT AND DISCUSSION
FIGURE 1 : INPUT
FIGURE 2: OUTPUT
CONCLUSION
It focused on how to use CNN models to predict
patterns of retinal disease using images from specific
datasets (trained datasets) and previous datasets. This
provides some of the following insights into the prediction
of retinal disease: The main advantage of the CNN
classification framework is the ability to automatically
classify images. Eye disorders are a major cause of
blindness and are often too late to correct. This study
provided an overview of how to detect retinal image
anomalies, including retinal image dataset collection,
preprocessing techniques, feature extraction techniques,
and classification schemes.
REFERENCES
[1] Choi JY, Yoo TK, Seo JG, Kwak J, Um TT, Rim TH (2017)
Multi-categorical deep learning Neural network to classify
retinal images: a pilot study employing small database. PLoS
One 12(11):e0187336
[2] Abramoff MD, Garvin MK, Sonka M (2010) Retinal
imaging and image analysis. IEEE Rev Biomed Eng. 3:169–
208
[3] Sinthanayothin C, Boyce JF, Cook HL, Williamson TH
(1999) Automated localisation of the optic disc, fovea, and
retinal blood vessels from digital color fundus images. Br J
Ophthalmol 83(8):902–910
[4] Hoover A, Goldbaum M (2003) Locating the optic nerve in
a retinal image using the fuzzy convergence of the blood
vessels. IEEE Trans Med Imag 22(8):951–958
[5] Sánchez CI, Hornero R, LopezMI, Poza J (2004) Retinal
image analysis to detect and quantify lesions associated with
diabetic retinopathy. In: 26thAnnual International
Conference of the IEEE engineering in medicine and biology
society, IEMBS’04, vol 1. IEEE, pp 1624–1627
[6] Mishra M, Nath MK, Dandapat S (2011) Glaucoma
detection from color fundus images. Int JComput Commun
Technol (IJCCT) 2(6):7–10
[7] Patton N, Aslam TM, MacGillivray T, Deary IJ, Dhillon B,
Eikelboom RH, Yogesan K, Constable IJ (2006) Retinal image
analysis: concepts, applications and potential. Prog Retinal
Eye Res 25(1):99–127
[8] Verma K, Deep P, Ramakrishnan A (2011) Detection and
classification of diabetic retinopathy using retinal images. In:
2011 Annual IEEE India conference (INDICON), pp 1–6. IEEE
[9] Priyadarshini BH Devi MR (2014) Analysis of retinal
blood vessels using image processing techniques. In: 2014
international conference on intelligent computing
applications (ICICA), pp. 244–248. IEEE 1280 K. Rajan and C.
Sreejith

RETINAL IMAGE CLASSIFICATION USING NEURAL NETWORK BASED ON A CNN METHODS

More Related Content

Similar to RETINAL IMAGE CLASSIFICATION USING NEURAL NETWORK BASED ON A CNN METHODS (20)

More from IRJET Journal (20)

Recently uploaded (20)

RETINAL IMAGE CLASSIFICATION USING NEURAL NETWORK BASED ON A CNN METHODS