1. PANIMALAR ENGINEERING COLLEGE
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
CS8811 PROJECT WORK
REVIEW NO:2
A Deep Learning Approach for the Detection of Diabetic Retinopathy
Using OCT Image
Guide Name: Sangeetha K
Team Members with Register number
Balaji R (211418104034)
Gokul N (211418104067)
BATCH NO: E-25
2. Abstract:
To classify normal and abnormal retinal images by using necessary
algorithms.
Apply Deep learning model to the field of medical diagnosis in order
to lessen the time and stress undergone by the ophthalmologist and
other members of the team in the screening, diagnosis and treatment
of diabetic retinopathy.
3. Abstract:
Proper detection of diabetic retinopathy in early stage is extremely
important to prevent complete blindness.
The Deep Learning models created by using the above neural
networks are capable of quantifying the features as micro aneurysms,
blood vessels, hemorrhages and fluid drip into different class of
categories.
The entire project has a user friendly which makes the identification
easy. Once the test image is uploaded, the interface will have buttons
in order to do the necessary transformation on the given image.
6. Existing system:
Fundus images are used for pre-processing
Pre-processing of image requires four steps:
i)Read image
ii)Resize image
iii) Noise removal
iv) segmentation
The existing system has less accuracy.
The resultant image will let us know whether the image is infected or
not. The stage of infection cannot be identified
7. Proposed system:
Optical coherence tomography(OCT) images are high resolution
images, contactless and non-destructive testing.
OCT is a cross sectional image of retina used
for classifying Diabetic retinopathy stage.
Accuracy is expected to increase with use of OCT image and
applying certain specific classification algorithm.
19. Literature survey:
TITLE 1 :
"Detection of Diabetic Retinopathy using Machine Learning" in International Research
Journal of Engineering and Technology (IRJET), 2020.
AUTHORS:
Aryan Kokane, Gourhari Sharma, Akash Raina, Shubham Narole, Prof. Pramila M.
Chawan,
DESCRIPTION:
The objective of this paper is to perform a survey of different literatures where a
comprehensive study on Diabetic Retinopathy (DR) is done and different Machine
learning techniques are used to detect DR. Diabetic Retinopathy (DR) is an eye
disease in humans with diabetes which may harm the retina of the eye and may
cause total visual impairment. Therefore it is critical to detect diabetic retinopathy
in the early phase to avoid blindness in humans. Our aim is to detect the presence
of diabetic retinopathy by applying machine learning classifying algorithms. Hence
we try and summarize the various models and techniques used along with
methodologies used by them and analyze the accuracies and results. It will give us
exactness of which algorithm will be appropriate and more accurate for prediction.
20. Literature survey:
TITLE 2 :
Diabetic retinopathy detection through deep learning techniques: A review" in
Informatics in Medicine Unlocked, 2020.
AUTHORS:
Wejdan L.Alyoubi, Wafaa M.Shalash, Maysoon F.Abulkhair
DESCRIPTION:
This paper used Convolutional neural networks are more widely used as a deep
learning method in detection of Diabetic retinopathy, the recent state-of-the-
art methods of Diabetic retinopathy color fundus images detection and
classification using deep learning techniques have been reviewed and
analyzed. Furthermore, the Diabetic retinopathy available datasets for the
color fundus retina have been reviewed. Difference challenging issues that
require more investigation are also discussed.
21. Literature survey:
TITLE 3 :
"Computer- Assisted Diagnosis for Diabetic Retinopathy Based on Fundus
Images Using Deep Convolutional Neural Network" in Mobile information
systems, 2019.
AUTHORS:
Yung-Hui Li , Nai-Ning Yeh, Shih-Jen Chen and Yu-Chien Chung
DESCRIPTION:
In the paper, a novel algorithm based on deep convolutional neural network
(DCNN). Unlike the traditional DCNN approach, we replace the commonly
used max- pooling layers with fractional max-pooling. Two of these DCNNs
with a different number of layers are trained to derive more discriminative
features for classification. After combining features from metadata of the
image and DCNNs, we train a support vector machine (SVM) classifier to
learn the underlying boundary of distributions of each class.
22. Literature survey:
• TITLE 4 :
• "Classification of Diabetic Retinopathy Images by Using Deep Learning Models" in
International Journal of Grid and Distributed Computing, 2018.
• AUTHORS:
• Suvajit Dutta, Bonthala CS Manideep, Syed Muzamil Basha, Ronnie D. Caytiles and N. Ch. S.
N. Iyengar
• DESCRIPTION:
• 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.
Click to add text
23. Literature survey:
TITLE 5 :
"Classifying Diabetic Retinopathy using Deep Learning Architecture" in International
Journal of Engineering Research & Technology (IJERT), 2016.
AUTHORS:
T Chandrakumar, R Kathirvel,
DESCRIPTION:
A proposed deep learning approach such as Deep Convolutional Neural Network(DCNN)
gives high accuracy in classification of these diseases through spatial analysis. A DCNN
is more complex architecture inferred more from human visual prospects. Amongst
other supervised algorithms involved, proposed solution is to find a better and
optimized way to classifying the fundus image with little pre-processing techniques.
Our proposed architecture deployed with dropout layer techniques yields around 94-
96 percent accuracy. Also, it tested with popular databases such as STARE, DRIVE,
kaggle fundus images datasets are available publicly.
24. Literature survey:
TITLE 6 :
Diabetic Retinopathy using Morphological operations and Machine Learning‖, IEEE
International Advance Computing Conference(IACC), (2015).
AUTHORS:
J.Lachure, A.V.Deorankar, S.Lachure, S.Gupta, R.Jadhav, ―
DESCRIPTION:
To develop this proposed system, a detection of red and bright lesions in digital fundus
photographs is needed. Micro-aneurysms are the first clinical sign of DR and it appear
small red dots on retinal fundus images. To detect retinal micro-aneurysms, retinal
fundus images are taken from Messidor, DB-rect dataset. After pre-processing,
morphological operations are performed to find micro-aneurysms and then features
are get extracted such as GLCM and Structural features for classification. In order to
classify the normal and DR images, different classes must be represented using
relevant and significant features. SVM gives better performance over KNN classifier.
25. Literature survey:
TITLE 7 :
SVM and Neural Network based Diagnosis of Diabetic Retinopathy‖,
International Journal of computer Application
AUTHORS:
R.Priya, P.Aruna
DESCRIPTION:
Two groups were identified, namely nonproliferative diabetic retinopathy (NPDR)
and proliferative diabetic retinopathy (PDR). In this paper, to diagnose diabetic
retinopathy, two models like Probabilistic Neural network (PNN) and Support
vector machine (SVM) are described and their performances are compared.
Experimental results show that PNN has an accuracy of 89.60% and SVM has an
accuracy of 97.608 %. This infers that the SVM model outperforms the other
model.
26. Literature survey:
TITLE 8 :
Identifying Abnormalities in the Retinal Images using SVM Classifiers‖,
International Journal of Computer Applications (0975- 8887), Volume 111 –
No.6,(2015).
AUTHORS:
S.Giraddi, J Pujari, S.Seeri
DESCRIPTION:
The aim of this paper is to develop and validate systems for detection of hard
exudates and classify the input image as normal or diseased one. The authors
have proposed and implemented novel method based on color and texture
features. Performance analysis of SVM and KNN classifiers is presented. Images
classified by these classifiers are validated by expert ophthalmologists.
27. Literature survey:
TITLE 9 :
Transformed Representations for Convolutional Neural Networks in Diabetic
Retinopathy Screening‖, Modern Artificial Intelligence for Health Analytic Papers
from the AAAI(2014).
AUTHORS:
G.Lim, M.L.Lee, Wynne Hsu
DESCRIPTION:
They demonstrate this functionality through pre-segmentation of input images
with a fast and robust but loose segmentation step, to obtain a set of candidate
objects. These objects then undergo a spatial transformation into a reduced
space, retaining but a compact high-level representation of their appearance.
Additional attributes may be abstracted as raw features that are incorporated
after the convolutional phase of the network. Finally, they compare its
performance against existing approaches on the challenging problem of
detecting lesions in retinal images.
28. Literature survey:
TITLE 10 :
Classification Algorithm of Retina Images of Diabetic patients Based on Exudates
Detection‖, 978-1-4673-2362-8/12, IEEE(2012)
AUTHORS:
Vesna Zeljkovic, Milena Bojic, Claude Tameze; Ventzeslav Valev
DESCRIPTION:
Automatic exudates detection and retina images classification would be helpful
for reducing diabetic retinopathy screening costs and encouraging regular
examinations. We proposed the automated algorithm that applies
mathematical modeling which enables light intensity levels emphasis, easier
exudates detection, efficient and correct classification of retina images. The
proposed algorithm is robust to various appearance changes of retinal fundus
images which are usually processed in clinical environments.
29. Model Features:
The project proposed is a deep learning model which is used for
identifying the level of infection in the human eye.
The model is well trained by datasets of the infected image which makes
the model more accurate and helps producing more perfect result.
The infected region along with the condition or level of the infection can
be identified using this model.
30. Image processing:
Optical coherence tomography(OCT) images are high resolution
images, contactless and non-destructive testing property.
Median filter The median filter is a non-linear digital filtering
technique, often used to remove noise from an image or signal.
Edge detection is used for image segmentation and data extraction in
areas such as image processing, computer vision, and machine
vision.
31. Statistical Data Extraction:
A collection of numerical data. The mathematical science dealing with
the collection, analysis, and interpretation of numerical data using the
theory of probability.
Standard deviation In statistics, the standard deviation is a measure of
the amount of variation or dispersion of a set of values.
When data is collected for that Statistic, it is compared with the
associated Threshold value.
32. Level Identification:
Supervised training Its use of labelled dataset to train algorithm that
to classify data or predict outcomes accurately.
Clusters are identified via similarity measures. These similarity
measures include distance, connectivity, and intensity.
Clustering is the task of dividing the population or data points into a
number of groups such that data points in the same groups are more
similar to other data points in the same group and dissimilar to the
data points in other groups.
33. Comparisons of Results:
Machine learning (ML) is the study of computer algorithms that improve
automatically through experience and by the use of data.
Machine learning algorithms build a model based on sample data, known
as "training data", in order to make predictions or decisions.
Neural networks are just one of many tools and approaches used in
machine learning algorithms.
34. Conclusion:
Thus from our existing system, we can be able to eliminate many
complexities faced by conventional detection systems. This system
heavily impacts and possibly reduces the possibility of lags and any
more inefficiencies that existed.
Automated systems for DR detection play an important role
in detection of the DR images due to its efficiency.
Most researchers have used the CNN for the classification and the
detection of the DR images due to its efficiency.
Statistical values can predict level of severity properly but in case of
noisy images the chances of getting poor data will lead
to lower accuracy. For yielding accurate result, selecting proper features
out of the image is also important.