A Novel Iris Cancer Detection using wavelet and wavelet packet Transform.

 

M. Saravanan, D. Kalpanapriya

VIT University, Vellore, Tamil nadu, India.

 

ABSTRACT:

This paper depicts simple and a helpful conclusion (normal or abnormal) of iris disease utilizing Wavelet Transform. The factual parameter like entropy extend are processed for iris image procured by Wavelet Transform and Wavelet Packet Transform. These parameters decide the variations from the norm showed in the iris image.In Image processing, we are introducing a novel approach to Wavelet transformation with respect to statistical parameter which are used to identify the tumor. The aim of this paper is to analyze this biomedical signal from the biomedical image of iris cancer and gather the information to diagnose, monitor, therapy control and evaluation.

 

 

INTRODUCTION:

The Iris are the hued part of the eye and it is comprised of two layers. The external "stroma" can be blue, hazel, green or cocoa. The back layer (the iris shade epithelium) is constantly chestnut. Tumors can develop inside and behind the iris. In spite of the fact that most iris tumors are growths or nevi, dangerous melanomas can likewise happen in the iris. The tumor may be seen by the patient, their family, or by the eye mind expert .A few people have bunches of spots on their Iris. Some of these pigmented spots have thickness, called Nevi. If the patient notification that one of their nevi has changed, broadened or is pulling on the understudy, they ought to see an eye administer to assessment and referral to an eye tumor expert.[7]Iris cancer is a very popular disease now days and it will mainly affect the child, so we need to identify the Detection of iris cancerIn Image processing some technics are used to identify the tumor. Helwan developed a new approach to identify the iris cancer using image filtering[10].

 

 

Eklundpresented the survey covers GPU increasing speed of fundamental image preparing operations (separating, interjection, histogram estimation also, separate changes), the most usually utilized calculations in restorative imaging (image enrollment, image segmentation, and image de-noising) and calculations that are particular to individual modalities .Abbasi introduced a programmed strategy for brain tumor discovery in 3D images.Haas found a new novel approach radio therapy images.Choi introduced a new technics used to detect the pulmonary nodule using genetic programming (GP)-based classifier genetic algorithm

 

The aim of our research work is to analyze this biomedical signal and to gather the information about diagnosis, monitoring, therapy ,control and evaluation. The nature of the biomedical signal determines diseases or defects in a biological system which causes alteration in normal physiological processes, leading to pathological processes. A pathological process is typically associated with signals that are different in some respects from the corresponding normal signals. We analyse Iris cancer images using transformation domain technique, specifically Discrete Wavelet Transform and Discrete wavelet packet transform which converts time domain data to frequency domain data.

 

System implementation

The Iris Image data for normal and cancer is obtained from Miles research and New York Eye Cancer Center through online .This image is pre-processed in the following few steps which are shown in Figure.1.shows that computation of Iris cancer detection by wavelet transform. Before applying the transform domain technics we are converting the coloured image into Grayscale image .The iris image (2Dsignal) is analyzed by two types of wavelet Transform . They are Discrete Wavelet Transform (DWT) and Discrete Wavelet Packet Transform(DWPT).

 

 

Figure.1. Flowchart of the detection of Iris cancer.

 

The figure.1.represents the flowchart of our work.it will illustrate the detection of iris cancer using image processing technics.

 

Wavelet Transform:

Wavelets are functions that decompose signals into frequency bands that possess MultiResolution Features .The wavelet transform computes the inner products of a signal with a family of wavelets. The wavelet transform tools are categorized into continuous wavelet tools and discrete wavelettools.

 

Discrete Wavelet Transform:

The Iris image using discrete wavelet transform was decomposed.Wavelets offer a simple and elegant framework for simultaneous timefrequencyand multiresolution analysis. In wavelet analysis, a function is decomposed into a superposition of dilated and scaled versions of a mother wavelet basis function. The decomposition highlights interesting features in the wavelet domain which are oftennot visible in the spatial domain. In DWT, an image is filtered into four subbands at each resolution and the low frequency subband is further filtered through an iterative process to provide the multiresolution representation. Here, we give an overview of the fast implementation of the discrete wavelet transform (DWT) [4].

 

Discrete Wavelet Packet Transform:

The Discrete wavelet packet transform is next level of the DWT.The Wavelet Packet Transform (WPT) adopts redundant basis functions and hence can provide an arbitrary time-frequency resolution. In this study, a WPT-based method is proposed for the damage assessment of structures. Dynamic signals measured from a structure are first decomposed into wavelet packet components. Component energies are then calculated and used as inputs into neural network models for damage assessment. Numerical simulations are performed on a three-span continuous bridge under impact excitation. The results show that the WPT-based component energies are good candidate indices that are sensitive to structural damage. These component energies can be used for various levels of damage assessment including identifying damage occurrence, location, and severity [3].The entropy computation also performed using DWPT only.

 

 

Figure.2.Wavelet Packet Decomposition Tree at Level 3

 

The figure.2.represents the decomposition level of wavelet packet transform.

 

A-approximation co-efficient of wavelets,S- (continuous in the variable(s)) offers an approximation to transform discrete (sampled) signals.

 

Denoise:

All digital images contain some degree of noise. Image de-noisingalgorithm attempts to remove this noise from the image. Ideally, the resulting de-noised image will not contain any noise or added artifacts. De-noising of natural images corrupted by Gaussian noise using wavelet techniques is very effective because of its ability to capture the energy of a signal in few energy transform values. The methodology of the discrete wavelet transform based image de-noising has the following three steps.

1     Transform the noisy image into orthogonal domain by discrete 2D wavelet transform.

2     Apply hard or soft thresholding the noisy detail coefficients of the wavelet transform.

3     Perform inverse discrete wavelet transform to obtain the de-noised image.

 

Here, the threshold plays an important role in the de-noising process. Finding an optimum threshold is a tedious process. A small threshold value will retain the noisy coefficients whereas a large threshold value leads to the loss of coefficients that carry image signal details. Normally, hard thresholding and soft thresholding techniques are used for such de-noising process. Hard thresholding is a keep or kill rule whereas soft thresholding shrinks the coefficients above the threshold in absolute value. It is a shrink or kill rule. The following are the methods of threshold selection for image de-noising based on wavelet transform, rest wavelet coefficients are very small. This algorithm offers the advantages of smoothness and adaptation.However, it exhibits visual artifacts [5].

 

Compression:

Compression is a main role of this work. Compression is classified into two types, lossy and lossless. The lossless compression recovers the origin data. In this paper, we approach a lossless compression which is the main goal of the algorithm. The image compression can be performed by coding methods, spatial domain technics and Transform domain technics. Coding method applied for raw images. Spatial domain technics will be applicable for both spatial and coding method, not applicable of direct gray value images. The transform domain technics are used for appropriate basic set images.

 

Global Threshold:

Basically two type of approaches are available in threshold- Global threshold and Local threshold. In this paper we approached the Global threshold technics also.Global threshold method is based on histogram of the image.The success of this technique is very strongly depends on how well the histogram can be partitioned.

The basic global threshold, (T)is calculated as following few steps:

1.      Select an initial estimate for T (typically the average grey level in the image)

2.      Segment the image using T to produce two groups of pixels: G1 consisting of pixels with grey levels more than T and G2 consisting pixels with grey levels ≤ T

3.      Compute the average grey levels of pixels in G1 to giveμ1 and G2 to give μ2

4.      Compute a new threshold value:

5.      Repeat steps 2 – 4 until the difference in T in successive iterations is less than a predefined limit T∞.

 

This algorithm works very well for finding thresholds when the histogram is suitable.

 

RESULTS:

The results were observed from the Normal and Iris cancer image (2D)signal through DWT method and DWPT method. Initially the images were de-noised. Then the de-noised images were applied in all the process. Next we appeal compression in different type of images. Compression technicssupplies the Residual in Normal iris image in which all pixels and white noise will present, but the Cancer image has pixel but no white noise. Absence of white noise clearly shown in the figure 4. The Global threshold will give the retained energy and number of zeros curve. The retained energy and number of zeros curves intersect in the iris cancer images. This absolute result will be useful for detection of iris cancer. DWT methodgives the histogramand the extracted frequency component has very low values for normal iris image(2D) signal , high values for the Iris cancer image (2D)signal and haveintermediate values for normal images (2D)signal. The same process is applied and analyzed through DWPT techniques.Entropy computation is one of the technic under DWPT which gives the higher value for Normal iris image and Lower value for Iris Cancer Image.

 

 

 

Simulation results of de-noised signal

 

Features	Original image	De-noised image
Normal iris image
Iris cancer image

Figure.3.Simulation results of de-noised images

 

The figure.3.represents the de-noised image of normal and iris cancer gray scale images. Our technics obtained the good results (i.e) the normal images are sharpen after de-noising but affected images are not sharpen after de-noising .

Simulation results of compression residuals and Global threshold

Features	Compressed (residuals) image	Global threshold
Normal iris image
Cancer iris image

Figure.4. Simulation results of compression (residuals image) and Global threshold

 

Thefigure.4. represents the compressed residuals images and global threshold results. Normal iris image is fully demolish the compression inner product pixels but in cancer imagesonly the cancer affected part demolish. This is one of the way to identify the affected iris cancer. In the normal iris images, retained energy and number of zeros are have same nearest value which has been shown in the graph , but these two curves never all cross anywhere but these two curves cross the global threshold point. After crossing the point thesetwo curves passes through the same values. In the iris cancer image, retained energy and number of zeros does not have the same nearest, but these two curves cross the global threshold point. Once it reach the global threshold point it will not passes the same value.

 

Simulation results of Histogram and Entropy

 

Features	Original image	Histogram	Entropy Tree
Normal iris image
Cancer iris image

Figure.5. Simulation results of histogram and entropy value

 

The figure.5.represents the histogram and entropy tree images of the normal and cancer iris images. DWT method supplies the histogram and the extracted frequency distribution component which has a very low values for normal iris images and high values for the cancer affected iris images. DWPT will give the entropy values in which the normal iris images have the high entropy which represents the present of low amount of white noise in the normal iris images and the low entropy value for cancer affected images represents the present of high amount of white noise in the cancer images.

 

Simulation results of computation of entropy

Features	Mathematical representation	Normal iris image value	Iris cancer
Entropy		<75000	>65000 Starting stage of cancer: (65000-75000)

 

Represents mathematical expression of entropy and statistical parameter of Normal and Iris cancer image.

 

CONCLUSION:

This paper introduced anovel approach for the detection of the iris tumor using Discrete wavelet transform and Discrete Wavelet Packet Transform under image processing. The aim of our research work is to analyze the biomedical signal and to gather the information about diagnosis, monitoring, therapy, control and evaluation. The paper provides the information of the tumorous iris image and normal iris image to processand to classify the regions including pupil for the affected images. Then the system has been tested for normal and cancer images which was obtained from Miles Research and New york cancer center. Further this study can be extend to prove that our effectivesystem can be implemented into the real life applications.

 

REFERENCE:

1.       Siegel. R. L, Miller, K. D, and Jemal. A ,Cancer statistics, A Cancer journal for clinicians, 66,2016, 7-30.

2.       Suraj, A, Francis. M, Kavya. T. S and Nirmal. T. M, Discrete wavelet transform based image fusion and de-noising in FPGA. Journal of Electrical Systems and Information Technology, 1, 2014, 72-81.

3.       Sun. Z, Chang. C. C, Structural damage assessment based on wavelet packet transform, Journal of structural engineering, 128, 2010, 1354-1361.

4.       Rajpoot. K, Rajpoot. N and Noble. J, Discrete wavelet diffusion for image denoising Image and Signal Processing,2008, 20-28.

5.       Mohideen.S.K., Perumal.S.A and Sathik.M.M, Image de-noising using discrete wavelet transform, International Journal of Computer Science and Network Security, 8, 2008, 213-216.

6.       Helwan. A, ITDS: Iris Tumor Detection System using Image Processing Techniques, International Journal ofScience Engineering Research, 5, 2014, 76-80.

7.       Eklund. A, Dufor, P, Forsberg. D and LaConte.S. M, Medical image processing on the GPU–Past, present and future,Medical image analysis, 17, 2013, 1073-1094.

8.       Abbasi. S and Tajeripour.F, Detection of brain tumor in 3D MRI images using local binary patterns and histogram orientation gradient, Neurocomputing, 219,2014, 526-535.

9.       Haas. O, Bueno.G, Spriestersbach. R, Himmler. H, Burnham.K and Mills, J,Imaging and control for adaptive radiotherapy, IFAC Proceedings , 38, 2005, 25-30.

10.     Choi. W. J and Choi.T. S, Genetic programming-based feature transform and classification for the automatic detection of pulmonary nodules on computed tomography images, Information Sciences, 212, 2012, 57-78.

11.     Hsun-Hsien Chang, Jose and M. F. Moura, Biomedical Signal Processing, , Biomedical engineering and design hand book,2 edition,vol.1,McGraw hill.20,chapter 22,559-579.

 

 

Accepted on 26.04.2018           © RJPT All right reserved

Research J. Pharm. and Tech 2018; 11(7): 2926-2931.