Vous êtes sur la page 1sur 43

Glaucoma Detection using Image Processing

Project Group Members:


Name of the Student Roll . No.
1) Pradnya Arun Shinde BEETB -- 249
2) Akshay Sanjay Manmode BEETB – 271
3) Mohini Balasaheb Kalekar BEETB – 275
4) Pratik Sudhakar Talware BEETB --278

Prof. Mrs. S.P. Jagtap Prof. A. B. Patil Dr. N. B. Chopade


Project Guide Project Co-Ordinator H.O.D (E&TC)
DEPARTMENT OF ELECTRONICS & TELECOMMUNICATION
B.E. (E&TC)
2017-18
INDEX
 Introduction
 Motivation
 Literature Survey
 Problem Statement
 Objectives
 Activity Chart
 Methodology Planned
 Block Diagram
 Hardware Specifications
 Software Specifications
 Circuit Diagram
 Result
 Flowchart
 Expected Conclusion
 References
INTRODUCTION
 Glaucoma is a disease that damages your eye’s optic nerve. It usually happens when fluid

builds up in the front part of your eye. That extra fluid increases the pressure in your eye,

damaging the optic nerve.

 It is non-curable, but if detected early, it can be avoided.

 Hence our project is on detection of Glaucoma disease detection using Image processing.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 1


 Glaucoma is an eye disease in which the pressure inside eyes increases enough to damage the optic
nerve and cause permanent vision loss.

Normal physiological parts of the eye fundus.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 4


 The increase in blood pressure happens when the passages that normally allow fluid in eyes to drain
become clogged or blocked.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 5


Cup to disc ratio calculation is an important part in Glaucoma screening.
CDR is a simply comparison of the cup (depressed area in the centre of the nerve) to the entire diameter
of the nerve.
The cup-to-disc ratio (CDR) expresses the proportion of the disc occupied by the cup and it is widely
accepted index for the assessment of glaucoma.
We segment the cup and disc in the fundus image and then calculate the cup and disc area and the cup to
disc ratio (CDR) and compare it to the database ratios.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 6


MOTIVATION
Statistics of no. of people with Glaucoma
70
No. of people worldwide (in million)

60

50

40
OAG
30 ACG
20

10

0
2010 2020
Year
OAG-Open Angle Glaucoma
ACG- Angle Closure Glaucoma
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 2
WOMEN
(WORLDWIDE) ASIANS

OAG 55% 47%

ACG 70% 87%

Link: http://bjo.bmj.com/content/80/5/389.short

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 3


LITERATURE SURVEY
Sr. No. Title of the Paper Year of Methodology Conclusion
publication/ used
Publisher

1. Comparative analysis of CDR detection The International Journal of In this proposed approach, This paper is presented and
Computer Science & preprocessing such as image evaluated for Glaucoma
for glaucoma diagnosis.
Applications,Volume2 no. 04, filtration, color contrast detection in patients, a global
June 2013 enhancement are performed leading cause for blindness.
which is followed by a
combined approach for image
segmentation and
classification using
binarization and
morphological operation.

2. Techniques of Glaucoma detection I.J. Image, graphics and In this paper the different In this paper the different
from color fundus images signal processing techniques techniques are discussed for
signal processing, 2012,
are discussed for detection glaucoma detection and
9, 44-51 published online and classification of classification. These methods
september 2012 in glaucoma. can be used in different
instruments
MECS.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 4


Diagnosis of glaucoma is instruments for easy
based on measurement of diagnosis of glaucoma
intraocular pressure by from the color fundus
tonometry, visual field images.
examination by perimetry
and measurement of cup
area to disc area ratio from
the color fundus images

3. Locating the Optic Nerve in a IEEE transaction on This method uses a novel This method evaluated
Retinal Image Using the Fuzzy medical imaging,Vol.22, algorithm we call fuzzy using 31 images of
Convergence of the Blood Vessels no.8, August 2003. convergence to determine the healthy retinas and 50
images of diseased
origination of the blood vessel
retinas, containing some
network. This method use the symptoms and achieved
convergence of the blood vessel 89% correct detection.
network as the primary feature
for detection, in conjunction
with the brightness of the nerve
as a secondary feature.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 5
A self-assessment reliability disc/cup boundary can
score is computed to be unclear at some
evaluate the quality of the sectors.
automated optic disc
segmentation

4. Glaucoma Detection From Submitted: March 18, 2012 The proposed system mainly In this study, a method
Fundus Image Using Opencv Accepted: April 14, 2012 consists of three different for detection of glaucoma
Published: December 15, stages. They are Region of based on the two features
2012 Interest (ROI) Extraction, CDR and ISNT ratio is
CDR Calculation and ISNT presented. The optic disc
Calculation. and optic cup from the
ROI image is extracted
and the area is calculated
using elliptical fitting
technique. Blood vessels
are enhanced using
matched filter response
and are extracted using
local entropy
thresholding technique.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 6


5. Glaucoma Diagnosis by Optic International Journal of In order to extract the optic This paper presented and
Cup to Disc Ratio Estimation Inventive Engineering and disc and cup, a region of evaluated for a more
Sciences (IJIES) ISSN: 2319– interest around the optic disc accurate estimation of
9598, Volume-3 Issue-5, must first be delineated, as neuro-retinal optic cup
April 2015 the optic disc generally based on multimodalities
occupies less than 5% of the including labeling
pixels in a typical retinal method, convex hull and
fundus image. While the disc ellipse fitting methods.
and cup extraction can be To determine the
performed on the entire performance of our
image, localizing the ROI approach, few retinal
would help to reduce the images were processed
computational cost as well as and their CDR was
improve segmentation calculated. Comparing
accuracy. with the clinical values,
the new approach
achieves a better CDR
value, which results to
more accurate Glaucoma
Diagnosis.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 12


PROBLEM STATEMENT

Design and implementation of a system which provides detection of a fatal


disease called Glaucoma using image processing and comparing the obtained
data with the pre-stored data and displaying the result on LCD with the help
of a PIC microcontroller.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 8


OBJECTIVE
 Obtain the image using fundus camera in JPEG format.

 Apply image preprocessing.

 Apply to Gabor filtering to image.

 Extract Region Of Interest (ROI)

 Optic Disc segmentation.

 Optic Cup segmentation.

 Calculate Cup-to-Disc Ratio (CDR)

 Display the output to 16x2 LCD.


1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 9
ACTIVITY CHART
Sr. Activity Jun Jul’ Aug’ Sep Oct’ Nov’ Dec’ Jan’ Feb’ Mar
No ’17 17 17 ’17 17 17 17 18 18 ’18
1. Problem
Statement
2. Literature Survey
3. Image Acquisition.
4. Image Pre -
Processing
5. Gabor Filtering.
6. ROI Extraction.
7. CDR Calculation.
8. Texting and
Experimentation.
9. Hardware Design
and
Implementation.
10. Report Writing.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 15
BLOCK DIAGRAM

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 10


Explanation of Block Diagram
1)Image acquisition:
 The retina can be observed and recorded using several methods including Fluorescein
Angiograms (FA), Transient Visual Evoked Potential (TVEP) or fundus camera.
 Fundus camera is preferred.
 It is color retinal fundus image.
 The images are stored in JPEG image format file.
 We apply this as a input to the pre-processing block.

Original Image
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 17
Explanation of Block Diagram (contd..)
2)Image Preprocessing:
• We can either apply basic filter techniques or we can use histogram equalization
techniques for the image processing.
• Adaptive histogram equalization(AHEM)is used to improve the contrast in image.
• It differs from ordinary histogram equalization.
• A variant of AHEM called CLAHE (contrast limited adaptive histogram
equalization) prevents the amplification of noise.
• We apply this JPEG image as a input for Gabor filtering.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 18


STEPS OF PRE PROCESSING
1. Image Resizing
2. Conversion of image into RGB channel
3. Adaptive Histogram Equalization

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 19


STEPS OF PREPROCESSING (contd…)

The image database is resized using imresize for removing redundant pixels so that the processing will
be fast but the pixels are removed sufficiently enough such that the output will not change.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 20


STEPS OF PREPROCESSING (contd…)

We have converted the image into its RGB channels.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 21


STEPS OF PREPROCESSING (contd…)

An image in its green channel is more sharp and thus we use green channel of an image for processing
further.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 22
STEPS OF PREPROCESSING (contd…)

We have performed histogram equalization of the image by using the MATLAB function histeq for
removing the noise in the image and increase the contrast of the image for better processing.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 23
STEPS OF PREPROCESSING (contd…)

Adaptive Histogram Equalization process

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 24


STEPS OF PREPROCESSING (contd…)

An image is always well illuminated at the centre and unevenly or not well illuminated as we go away
from the centre. Therefore, to evenly illuminate the image we use adapthisteq function which performs
Adaptive Histogram Equalization of the image and evenly illuminates it.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 25
Explanation of Block Diagram (contd..)
3) Gabor Filtering.
 In image processing, a Gabor filter, is a linear filter used for edge detection.
 Frequency and orientation representations of Gabor filters are similar to those of the
human visual system, and they have been found to be particularly appropriate for texture
representation and discrimination.
 Gabor wavelets can be tuned for specific frequencies and orientations which is useful for
blood vessels.
 They act as low level oriented edge discriminators and also filter out the background noise
of the image.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 26


Results of Gabor Filtering

Fig. (a) Fig. (b)


The fig. (a) shows the first step of Gabor Filtering i.e. Image enhancement. This is done to enhance the image
more to obtain its features by transforming the values using contrast-limited adaptive histogram equalization
(CLAHE). The fig.(b) shows the adjusted version of the enhanced image. The function imadjust is used to do
so. Using this, the low and high intensities of the image are adjusted such that to adjust the image intensity
values of the image. This increases the contrast of the output image.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 27
Results of Gabor Filtering (contd…)

The next step involves actual Gabor Filtering for which we have written a user defined function gab_roi. In
this function, we are using Sobel Edge Detection by using the function edge(image, 'sobel') which will
detect the edges of our image for manual tracing of the required portion.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 28
Explanation of Block Diagram (contd..)
4) Region of Interest (ROI):
 A region of interest (ROI) is a portion of an image that you want to filter or
perform some other operation on.
 You define an ROI by creating a binary mask, which is a binary image that is
the same size as the image you want to process.
 In the mask image, the pixels that define the ROI are set to 1 and all other
pixels set to 0.
 The fundus image is subdivided into number of regions, and an approximate
ROI Centre is selected based on the region containing the highest number of
pre-selected pixels

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 29


Region of Interest (ROI) (contd…)

The above is the manually traced portion of our image on which we are going to work. The interactive
session between the user and the image is done by using the function imfreehand that enables us to trace
the required portion for our operation and work on only that portion. This portion is called Region Of
Interest (ROI). After this, a mask of the ROI is created using the function createMask which will create a
white mask of the ROI for segmentation of the ROI from the adjusted image.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 30
Fig. (a) Fig. (b)
The fig(a) shows the white mask that is acquired using the createMask. The fig(b) is the required portion from
the adjusted image that is needed for further operation. The image is the compliment of XOR of the adjusted
image and the white mask shown in fig(a). This is the portion of the adjusted image that is required for doing disc
and cup segmentation.
1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 31
Explanation of Block Diagram (contd..)
5) Disc Extraction:
 Feature extraction: Gabor filter is used.
 Gabor wavelets can be tuned for specific frequencies and orientations
which is useful for blood vessels.
 It also filters out the background noise of the image reduction.
 Gabor wavelets can be tuned for specific frequencies and orientations
which is useful for enhancement of blood vessels by which we get proper
shape of optical disc.
 Morphological operation: Here erosion is used to remove extra pixels to
the boundaries of objects in an image by using structuring element to
reshape the obtained disc boundary and get proper disc shape.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 32


Explanation of Block Diagram (contd..)

6) Cup Extraction:
 We can use thresholding or binarization for optic cup segmentation
process.
 This process will convert the given image into a threshold or binarized
image where we can easily get our optic cup.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 33


Explanation of Block Diagram (contd..)
7)CDR calculation:
 After obtaining the disc and cup, various features can be computed. We
follow the clinical convention to compute the CDR.
 As mentioned in the introduction, CDR is an important Indicator for
glaucoma screening computed as,
CDR = Area of Cup/Area of Disc
 The computed CDR is used for glaucoma screening.
 CDR value exceeds 0.6 shall be recommended for further analysis of a
patient to the ophthalmologist.
 When CDR is greater than a threshold, it is glaucomatous, otherwise it
will be considered as a healthy one.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 34


HARDWARE SPECIFICATIONS
1. PIC16F877A:
Sr. No. Key Features Specifications

1. Operating frequency DC-20MHz

2. Resets(and Delays) POR,BOR(PWRT, OST)

3. Flash Program Memory (14-bit words) 8K

4. Data Memory (bytes) 368

5. EEPROM Data Memory (bytes) 256

6. Interrupts 15

7. I/O Ports Ports A,B,C,D,E

8. Timers 3

9. Compare/Capture/PWM modules 2

10. Serial Communication MSSP/USART

11. Parallel Communication PSP

12. 10-bit Analog-to-Digital Module 8-input channels

13. Analog Comparators 2

14. Instruction Sets 35 instructions

15. Packages 40-pins PDIP


44-pins PLCC
44-pins 1QFP
44-pins QFN

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 11


HARDWARE SPECIFICATIONS (contd…)
2. JHD162A Series:
Parameter: V∞=5.0V±10%, Vss=0V, Ta=25°C.
Parameters Symbols Testing Standard Vaues
Criteria Min Typ Max Unit

Supply Voltage VDD-VSS - 4.5 5.0 5.5 V

Input high voltage VIH - 2.2 - VDD V

Input low voltage VIL - -0.3 - 0.6 V

Output low voltage VOL IOL=1.2mA - - 0.4 V

Output high voltage VOH -IOH= 02mA 2.4 - - V

Operating Voltage IDD VDD=5.0V - 1.5 3.0 mA

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 12


SOFTWARE SPECIFICATIONS
MATLAB:
 MATLAB version R2012A.
 A high-performance language for technical computing.
 It integrates computation, visualization, and programming in an easy-to-use environment.
 Typical uses include:
 Math and computation
 Algorithm development
 Modeling, simulation, and prototyping
 Data analysis, exploration, and visualization
 Scientific and engineering graphics.
 Application development, including Graphical User Interface building
 Many technical computing problems are solved
 Small span of time to write a program

NI MULTISIM:
 It is an electronic schematic capture and simulation program.
 Designed for schematic entry, simulation and feeding to downstage steps, s uch as PCB layout.
 Multisim version 7.8.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 13


CIRCUIT DIAGRAM

Input: 230V, 50Hz.


Expected Output: 5V, 300mA

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 38


RESULTS

VALUES:
• Ideal:
1) Across Rectifier:15V
2) Across Output: 5V
• Practical:
1) Across Rectifier: 14.583V
2) Across Output: 5.296V
Hence we obtained the expected results.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 39


FLOWCHART

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 14


EXPECTED CONCLUSION
1. Effectively process the various color fundus images using Gabor Filtering and Region Of Interest
(ROI).
2. For determination of the presence of the disease, we will calculate the Cup to Disc Ratio (CDR),
which will then be compared to the standard ratio values:
 CDR<0.4, then it is normal.
 CDR>0.4 but <0.6, then it is moderate glaucoma.
 CDR>0.6, then it is severe glaucoma.
and will detect the Glaucoma disease, if any, with maximum accuracy.
3. We will display the results of the stage of the disease on LCD using microcontroller.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 15


REFERENCES
Papers
(1) Adam Hoover and Michael Goldbaum” Locating the Optic Nerve in a Retinal Image Using the Fuzzy Convergence of
the Blood Vessels” IEEE transactions on medical imaging, vol. 22, no. 8, august 2003.
(2) Malaya Kumar Nath and Samarendra Dandapat” Techniques of Glaucoma Detection From Color Fundus Images: A
Review” I. J. Image, Graphics and Signal Processing, 2012, 9, 44-51 Published Online September 2012 in MECS.
(3) S.Chandrika1, K.Nirmala2 1&2 Department of ECE GKM College of Engineering & Technology, Anna University New
Perungalathur, Chennai 1”comparative analysis of CDR detection for glaucoma diagnosis” Volume 2, No. 04, June 2013
ISSN – 2278-1080 The International Journal of Computer Science & Applications.

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 16


Books
• Digital image processing for Ophthalmology: detection of the optic nerve head, Xiaolu Zhu published by
Morgan and Claypool.

 Image analysis and modeling in Ophthalmology, eddie Y.K.ng, published by CRC press.

Websites

 http://www.optic-disc.org/

 http://www.nei.nih.gov/

1/14/2020 Pimpri Chinchwad College of Engg, Nigdi-44 17

Vous aimerez peut-être aussi