The e-ophtha-EX (Exudates) Images of healthy patients with no

database consists of 89 color fundus images of which 84 contain at least mild
non-proliferative signs (Microaneurysms) of the diabetic retinopathy, and 5 are
considered as normal as shown in the Fig. 5 which do not contain any signs of
the diabetic.

  The DRIVE dataset
consists of 40 images which has been divided into a training and a test set,
both containing 20 images as shown in Fig. 6

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

  The 1200 eye fundus
color numerical images of the posterior pole for the MESSIDOR database were
acquired by 3 ophthalmologic departments using a color video 3CCD camera. The
images were captured using 8 bits per color plane at 1440*960, 2240*1488 or
2304*1536 pixels. 800 images were acquired with pupil dilation and 400 without
dilation. The 1200 images are packaged in 3 sets, one per ophthalmologic
department. Each set is divided into 4 zipped sub sets containing each 100
images in TIFF format and an Excel file with medical diagnoses for each image.

  E-ophtha is a
database of color fundus images especially designed for scientific research in
Diabetic Retinopathy (DR) as shown in the Fig. 8. The database is made of
retinal images with different types of lesions (exudates and microaneurysms)
manually annotated by ophthalmology experts. E-ophtha is made of two sub
databases named e-ophtha-MA (Microaneurysms), and e-ophtha-EX (Exudates) Images
of healthy patients with no lesion are also provided in the two databases.


Image Preprocessing:

  Several preprocessing
methods are used to implement the initial processing, namely as gamma
correction, green channel extraction, histogram equalization, etc. In the
proposed method the techniques used are Gaussian filter and CLAHE.

  Gaussian filtering is
used to blur images and remove noise and detail. The Standard deviation of the
Gaussian function plays an important role in its behavior. The values located
between +/- ? account for 68% of the set, while two standard deviations from
the mean (blue and brown) account for 95%, and three standard deviations (blue,
brown and green) account for 99.7%. This is very important when designing a
Gaussian kernel of fixed length. The Gaussian function is used in numerous
research areas:

It defines a probability distribution for noise or data.

It is a smoothing operator.

It is used in mathematics.


  CLAHE operates on
small regions in the image, called tiles, rather than the entire image that is
to be processed. Each tile’s contrast is enhanced, so that the histogram of the
output region approximately matches the histogram specified by the Distribution
parameter. The neighboring tiles are then combined using bilinear interpolation
to eliminate artificially induced boundaries. The contrast, especially in
homogeneous areas, can be limited to avoid amplifying any noise that might be
present in the image.


J=adapthisteq(I,param1,val1,param2,val2…) specifies any of the
additional parameter/value pairs listed in the following table. Parameter names
can be abbreviated, and case does not matter. CLAHE is an adaptive contrast
enhancement method. It is based on AHE, where the histogram is calculated from
the contextual region of the pixel.


 Wavelet Transformation:

  In continuous wavelet
transforms, a given signal of finite energy is projected on a continuous family
of frequency bands (or similar subspaces of the Lp function space L2(R) ). For
instance the signal may be represented on every frequency band of the form f,
2f for all positive frequencies f > 0. Then, the original signal can be
reconstructed by a suitable integration over all the resulting frequency


  In discrete wavelet
transforms, it is computationally impossible to analyze a signal using all
wavelet coefficients, so one may wonder if it is sufficient to pick a discrete
subset of the upper half plane to be able to reconstruct a signal from the
corresponding wavelet coefficients. One such system is the affine system for
some real parameters a > 1,   b > 0.

  Among the various
wavelets, Mathieu wavelet is used in the proposed method. The Mathieu equation
is a linear second-order differential equation with periodic coefficients.
“Mathieu functions are applicable to a wide variety of physical phenomena,
e.g., diffraction, amplitude distortion. Mathieu’s equation is related to the
wave equation for the elliptic cylinder. The Mathieu equation is given by,

 +(a-2q cos 2?)y=0

   Mathieu wavelets can
be derived from the low pass reconstruction filter by the cascade algorithm.
Infinite Impulse Response filters (IIR filter) should be use since Mathieu
wavelet has no compact support.


§  Even periodic

cer(?,q) =? Ar,m cos m? for a =ar(q)


Odd periodic solution

ser(?,q) =? Ar,m sin m? for a =ar(q)


which is known as the dilation or refinement equation, is the
chief relation determining a Multiresolution Analysis (MRA).



 is the transfer
function of the smoothing filter.



is the transfer function of the detail filter.


The transfer function of the “detail filter” of a
Mathieu wavelet is,

 Gv(? =

  The transfer function
of the “smoothing filter” of a Mathieu wavelet is,


Feature Extraction:

  Statistical features
are extracted from the transformed image and their correlation with the
detection of microaneurysms automatically was studied. Various features
relating size and shape are extracted in feature. Also many studies state that
gabor features are also used in analysis and detection of diabetic retinopathy
at its early stage of occurrence. Hence extraction of the features from the
transformed image should be made in a relative manner.



  Once microaneurysms are detected they are classified on the
basics of the number of microaneurysms present in the retinal image as shown in
Table 1.


1 Criteria Used for Grading the Diabetic Retinopathy

DR Stage

No. of Microaneurysms

Grade 0(no DR)

MA =0

Grade 1(mild)


Grade 2 (moderate)


Grade 3(severe)




  MATLAB is a high-level language and
interactive environment for numerical computation, visualization, and
programming. Using MATLAB, you can analyze data, develop algorithms, and create
models and applications. The language, tools, and built-in math functions
enable you to explore multiple approaches and reach a solution faster than with
spreadsheets or traditional programming languages, such as C/C++ or Java. You
can use MATLAB for a range of applications, including signal processing,etc.,
More than a million engineers and scientists in industry and academic use
MATLAB, the language of technical computing.



  The comparison of the
features between the microaneurysms and non microaneurysms in terms of
mathematical values which are features extracted from the retinal image which
are taken from the publically available database DIARETDB1, such the first
order statistics and DWT features is listed below as shown in the Table 2.


Table 2
Performance of Existing Method



Non MA

First Order Statistics




Standard deviation



DWT features





















  The GUI display model of detecting the
microaneurysms and classifying them in an easy manner. This model contains of
two sections namely the following:

Training Section

Testing Section



                                        Fig. 9 Input Image for Training                    Fig. 10 Preprocessed Image

  Fig. 9 is the initially used image for
training the network. Fig. 10 shows the output of pre-processed image that is
obtained by using CLAHE filtering, also this filter is used in many other
images for preprocessing such as brain waves, etc. It improves the quality of
an image by enhancing the contrast.



                    Fig. 11 Mathieu
decompossed Image         Fig. 12 Mathieu
reconstructed Image

  The Mathieu wavelet decomposed image by using
the CLAHE filter in preprocessing is shown in the Fig. 11. It is a segmentation
transformation method for the detection of microaneurysms.

  The image is reconstructed from the obtained
decomposed image, as shown in the Fig. 12.


       Fig. 13 Input Test Image

  Fig. 13 is the test image used in the
analsis, any kind of image can be used for testing in the process.




                   Fig. 14 a)Grey Level
Image        b) Preprocessed Image

  The test image is moved to the next
processing level and they are converted as the Gray level image and next
processing level is the preprocessing and the output images are resulted as in
the Fig. 14  a) & b) respectively.





                                         Fig. 15 Decomposed Image              Fig. 16 Reconstructed Image  

Decomposed image from the test set is 
shown in the Fig. 15. The Fig. 16 shows the reconstructed image that is
chosen from the test set. The output is taken for 20 images from 4 different
databases, and the final network performance is obtained to be of merely 100%
of Accuracy and Specificity. The output window that shows the classification of
diabetic retinopathy is shown further.