DIAGNOSIS OF SKIN CANCER USING MORPHOLOGY IN IMAGE

PROCESSING

ABSTRACT:

Cancer, a dreadful disease characterized by uncontrolled growth and spread

of abnormal cells cause death worldwide. The National Cancer Institute estimated
that the cost of melanoma was $2.36 billion in 2010 and will continue to increase
in upcoming years. Skin cancer is the deadliest form of cancers in humans. Skin
cancer is of two types namely: malignant melanoma and non-malignant melanoma.
Melanoma can be cured completely if it is detected early. Only an expert
dermatologist can classify which one is malignant. And which one is non-
malignant.
A recent and novel method of melanoma detection is the "ugly duckling
sign” It is simple and highly effective in detecting melanoma. Extra care and
caution should be rendered when examining such individuals, as they might have
multiple melanomas and unusual mole. The borders of these melanomas are often
indistinct, making visual identification without a dermatoscope very difficult.
The dermoscopy image of skin cancer is taken and it is subjected to various
pre-processing and image filtering. The cancer affected region is separated from
the healthy skin using Segmentation.
Medical images play vital role in assisting health care providers in proper
diagnosis and correct treatment. Availability of pathologist in the remote areas is
less, to study the pathology images such as cancer images. Digital image
processing techniques can identify the features more accurately and provide the
appropriate status on disease. Consequently the use of computer-aided systems
becomes very essential to overcome these limitations.The proposed scheme is
using Wavelet Weiner filtering for cancer skin removal, morphological processing
algorithms like boundary extraction, thinning, thickening and skeleton and
thresholding based segmentation using Otsu’s method. Dermatologists use the
ABCDE rule to help people spot the signs of melanoma on their skin.
Total body photography, which involves photographic documentation of as
much body surface as possible, is often used during follow-up of high-risk patients.
The technique has been reported to enable early detection and provides a cost-
effective approach (being possible with the use of any digital camera), but its
efficacy has been questioned due to its inability to detect macroscopic changes.The
diagnosis method should be used in conjunction with (and not as a replacement
for) dermoscopic imaging, with a combination of both methods appearing to give
extremely high rates of detection.

Keywords— Melanoma; Dermoscopy; Curvelet Weiner filtering; Boundary

extraction; Thinning; Thickening; Skeletons; Thresholding; Segmentation
Literature survey

UzmaBanoAnsari1(2017) Skin cancer is the most common cause of death

amongst humans. Skin cancer is abnormal growth of skin cells most often develops
on body exposed to the sunlight, but can occur anywhere on the body. Most of the
skin cancers are curable at early stages. So an early and fast detection of skin
cancer can save the patient’s life. With the new technology, early detection of skin
cancer is possible at initial stage. Formal method for diagnosis skin cancer
detection is Biopsy method [1]. It is done by removing skin cells and that sample
goes to various laboratory testing. It is painful and time consuming process. We
have proposed skin cancer detection system using svm for early detection of skin
cancer disease. It is more advantageous to patients. The diagnosing methodology
uses Image processing methods and Support Vector Machine (SVM) algorithm.
The dermoscopy image of skin cancer is taken and it goes under various pre-
processing technique for noise removal and image enhancement. Then the image is
undergone to segmentation using Thresholding method. Some features of image
have to be extracted using GLCM methodology. These features are given as the
input to classifier. Support vector Machine (SVM) is used for classification
purpose. It classifies the given image into cancerous or non-cancerous

ChandrahasaM ,VarunVadigeri and Dixit Salecha(2016) Smartphones

are playing major role in e-health in such a way that m-health is playing a
significant role in healthcare industry. Image processing techniques are
instrumental in healthcare industry to detect abnormalities in human body. Skin
cancer (Melanoma) is one of the most deadly cancers, but when diagnosed early, it
can be cured. Reports tell that more than million deaths occur due to Skin cancer
itself. This paper speaks about how skin cancer can be detected in early stages
using smartphone application by analyzing properties of the cancer, Asymmetry,
Border, Color variation, Diameter and Expansion(ABCDE).These properties are
analyzed using different image processing techniques like Grey scale conversion,
Segmentation, contour tracing and histogram analysis.

Rafealgracia (2012) Computerized analysis of pigmented skin lesions

(PSLs) is an active area of research that dates back over 25years. One of its main
goals is to develop reliable automatic instruments for recognizing skin cancer from
images acquired in vivo. This paper presents a review of this research applied to
microscopic (dermoscopic) and macroscopic (clinical) images of PSLs. The review
aims to: (1) provide an extensive introduction to and clarify ambiguities in the
terminology used in the literature and (2) categorize and group together relevant
references so as to simplify literature searches on a specific sub-topic. Methods and
material: The existing literature was classified according to the nature of
publication (clinical or computer vision articles) and differentiating between
individual and multiple PSL image analysis. We also emphasize the importance of
the difference in content between dermoscopic and clinical images. Results:
Various approaches for implementing PSL computer-aided diagnosis systems and
their standard workflow components are reviewed and summary tables provided.
An extended categorization of PSL feature descriptors is also proposed, associating
them with the specific methods for diagnosing melanoma, separating images of the
two modalities and discriminating references according to our classification of the
literature. Conclusions: There is a large discrepancy in the number of articles
published on individual and multiple PSL image analysis and a scarcity of reported
material on the automation of lesion change detection. At present, computer-aided
diagnosis systems based on individual PSL image analysis cannot yet be used to
provide the best diagnostic results. Furthermore, the absence of benchmark
datasets for standardized algorithm evaluation is a barrier to a more dynamic
development of this research area.

JosuéÁlvarez-Borrego(2015) In this paper a new methodology for the

diagnosing of skin cancer on images of dermatologic spots using image processing
is presented. Currently skin cancer is one of the most frequent diseases in humans.
This methodology is based on Fourier spectral analysis by using filters such as the
classic, inverse and k-law nonlinear. The sample images were obtained by a
medical specialist and a new spectral technique is developed to obtain a
quantitative measurement of the complex pattern found in cancerous skin spots.
Finally a spectral index is calculated to obtain a range of spectral indices defined
for skin cancer. Our results show a confidence level of 95.4%.

M.M. Rahman, P. Bhattacharya(2009) This paper presents an integrated

and interactive decision support system for the automated melanoma recognition of
the dermoscopic images based on image retrieval by content and multiple expert
fusion. In this context, the ultimate aim is to support the decision making by
retrieving and displaying the rel-evant past cases as well as predicting the image
categories (e.g., melanoma, benign and dysplastic nevi)by combining outputs from
different classifiers. However, the most challenging aspect in this domain is to
detect the lesion from the healthy background skin and extract the lesion-specific
local image fea-tures. A thresholding-based segmentation method is applied on the
intensity images generated from two different schemes to detect the lesion.

FatemehTorkashvand, MehdiFartash(2015) The image segmentation is

one of the most important topics in the image processing that plays a key role in
the image analysis. This study presents a new method to segregate the skin lesions
through level set concept with regard to the level of homogeneity and Markov
random field. A cluster property with local intensities was extracted from the
image during the smoothing and level set phases based on the intensity of
heterogeneity and a local clustering function was defined for the intensities of each
pixel adjacent to each point. Then, an image is segmented by Markov random field
and allocation of each pixel of the image to existing classes that would finally lead
to the lesion segregation from the healthy skin. Our method was tested on 200
dermoscopic images taken from the PH database in various color spaces.
According to the results and comparison with the results of clinical diagnosis by
dermatologists, the proposed method had 94% accuracy in green channel, which
indicates a better performance than other color spaces.
SYSTEM REQUIREMENT:
HARDWARE REQUIREMENTS:

Processor : PENTIUM IV
Ram : 1 GB SD RAM
Monitor : 15” COLOR
Hard Disk : 80 GB
Keyboard : STANDARD 102 KEYS
Mouse : 3 BUTTONS

SOFTWARE CONFIGURATION:

Operating System : Windows

Environment : MATLAB
Matlab : Version 12a
INTRODUCTION:
Cell morphology has become a standard theory for computerized cell image
processing and pattern recognition. Morphological cell analysis is a key issue for
abnormality identification and classification, early cancer detection, and dynamic
changes analysis under specific environmental stress. As more and more imaging
data becomes available, innovative software algorithms for image processing and
analysis will be a critical need for effective use of information presented by
medical images.

Melanomas can occur anywhere on the body, but are more likely to start in certain
locations. The trunk (chest and back) is the most common site in men. In women,
the legs are the most common site. The neck and face are other common places for
melanoma to start.

STEPS IN SKIN CANCER DETECTION SYSTEM

This section has a review on the diagnosis of skin cancer detection systems.
The common approach of designing is divided into six steps as illustrated in Fig 1

Image Acquisition
 The database images we use contain digital images taken by means of a
digital camera.
 These images are fed into a computer system for further processing. The
images mainly dealt for the research are RGB images. Since color is a
powerful descriptor the RGB images are considered for the paper. The
database images are obtained from different sources and the size of the
images is non-standard.
 Fig: Input Image

Image preprocessing
Preprocessing as the fundamental stage of detection system helps to enhance
the quality of an image by removing hairs, noise and air bubbles on the skin. The
enhanced image is used for feeding the next step. In preprocessing of an image,
there are many existing techniques which can be classified into two groups; binary
and gray color images. The common images chosen for research here are color
images.

Fig: preprocessed image

Wiener filter purposed to reduce the amount of noise presented in a signal by
comparison with an estimation of the desired noiseless signal. It work based on
neighbor hooding pixels of size mby-n to estimate the local image mean and
standard deviation. The additive noise (Gaussian white noise) power is assumed to
be noise for suppressing the noise in theimage by removing the outliers that are the
extreme pixel values.

Fig: Architecture

Wiener filtering

Curvelet Wiener filtering executes an optimal tradeoff between inverse filtering

and noise smoothing. It removes the additive noise and inverts the blurring
simultaneously.

Fig: Filter applied Image

 Wiener filter estimates the local mean and variance around each pixel.
Where is the N-by-M local neighborhood of each pixel in the image, then
creates a pixel-wise wiener filter using these estimates, Where the noise variance
is not given, then the average of all the local estimated variances.
The enhancement of skin cancer images is performed by using Curvelet
algorithm. They have performed both the enhancement of images and reduction of
noise. The benefit of Curvelet based technique is that it is receptive to boundaries
at various angles and better extracts the high pass or finest details of object
contours at multiple scales through multiple nonzero coefficients. A local contrast
enhancement method for RGB images utilizes morphological filtering to obtain
the scale specific dark and bright features from the input image.

PROPOSED METHOD
In our methodology, the image is first enhanced by obtaining the highest
frequency components from its Curvelet transform and then add it to the original
image, in order to sharpen the edge detail. Subsequently the sharpened image is
subjected to morphological processing and thresholding to get a binary image,
from which boundaries are extracted after morphological processing. In the end, an
Otsu algorithm is applied to get normal skin and the cancerous skin. I thus
propose a computerised solution for replacing the clinical calculations by feature
exatraction.

Our procedure involves the following steps:

1. For any segmentation strategy, noise removal is a must, a priori, lest one
may get a lot of false edges. Our method starts with the removal of unwanted
 particles or noise present in the image (I), through the use of Weiner filter to get
IW. The latter is useful in the situations where the purpose is to reduce noise but
preserve the edges. Wiener filter is statistical in nature as it adopts a least square
(LS) approach for signal recovery in the presence of noise. It is very effective in
eliminating both the additive noise and blur which are usually competing against
each other.

2.A Forward Discrete Curvelet Transform (FDCT) is applied to the input image to
get the finest detailed coefficients. The FDCT is a multi-dimensional transform in
the sense that not only linear contours but also the curvy edges of the contained
objects can be captured through its use. Hence, the Curvelet transform captures
the structural activity along the radial wedges in the frequency domain and has a
very high directional sensitivity. It captures singularities with very few
coefficients in a non-adaptive manner. The edge and singularity details are
processed to extract the feature points.

3.The obtained high-pass image (IHP) is added to IW and we get an enhanced SEM
image (Ie). The image would now have stronger edges than the original and
would perform better in lending edge details to the segmentation step.

4.The mask is further refined via Mathematical Morphology (MM) processing,

getting (IM), in order to further highlight the image boundaries. The segmented
image (IS) is formed by superimposing the mask (IM) on the image IE and the
regions are separated by setting all the pixels to 1 that belong to the set of the
segmentation boundaries.
(a). Boundary extraction:

Fig: Binary Image

The boundary of a set A, denoted as β(A), can be obtained by first eroding A by B
and then performing the set difference between A and its erosion as follows:
β ( A) = A− ( A B)
where
B is a suitable structuring element.
A binary image Boundary extracted using a 3x3 structuring element of ones
Size of structuring element defines the boundary being 1 pixel thick.
(b) Thinning
The thinning of a set A by a structuring element B is defined in terms of the
hit-or-miss transform:

AⓩB = A− ( A⊗B) = A∩( A⊗B)c

So far, we were interested only in pattern matching with the structuring elements,
so no background operation is required in the hit-or-miss transform. A more useful
expression for thinning Asymmetrically is based on a sequence of structuring
elements:
{ B} = { B1, B2 ,...Bn}
The process is to thin A by one pass with B1, then thin the result with one
pass of
B2 d tillAi
Set Sequence of rotated SEs
B2, and so on, until is thinned with one pass of Bn.
The entire process is repeated until no further changes occur.

(c)Thickening:
Depending on the nature of A, the thickening procedure may result in
disconnected points. Therefore, this method is usually followed by post-processing
to remove disconnected points.
(d). Skeleton:

A skeleton S(A) of a set A can be viewed as:

a) If z is a point of S(A) and (D)z is the largest disk centered at z and
contained in A, one cannot find a larger disk (not necessarily centered at z)
containing (D)z and included in A. The disk (D)ziscalled a maximum disk.
b) The disk (D)ztouches the boundary of A at two or more different places.
Segmentation

The segmentation is the most important stage for analyzing image properly
since it affects the accuracy of the subsequent steps. However, proper segmentation
is difficult because of the great verities of the cancer shapes, sizes, and colors
along with different skin types and textures. In addition, some cancer has irregular
boundaries and in some cases there is smooth transition between the lesion and the
skin.
To address this problem, several algorithms have been proposed. In this
thesis I have discussed one method of thresholding using Otsu’s method.
Otsu’s thresholding is one of the oldest methods in the image segmentation. It is
treated as a statistical method according to probabilistic implemented. It should be
noted that the method Otsu is one of the best ways automatic threshold level. The
basic principle in Otsu method is divided into two categories, such as foreground
and the background. The threshold is obtained by finding automatic maximum
contrast between the layers.

Fig: Segmented Image

Otsu based on the pixel density contrast. Otsu method in the threshold level
iterating through all possible limit values and calculate the behavior of the
spreading the pixels levels around each side of the threshold, any pixel may fall
either in the foreground or background.
Feature extraction

Feature extraction is extracting the most reliable, measurable and sensitive

features to be supplied to the classifiers. The most well-known models of feature
extraction in skin cancer images are Pattern analysis, ABCDE-rule of
dermatoscopy.

Dermatologists use the ABCDE rule to help people spot the signs of melanoma on
their skin:
 o Asymmetry
o Border
o Color
o Diameter
o Evolving

SAMPLE CODING
 close all
 clear all
 clc
 %%
 globalInput_Image
 global EN_IP


 % Browse input image from Data base



 [filename pathname id] = uigetfile({'* .jpg'},'Select input image',[cd

'\images']);
 if ~id,return,end


 Input_Image = double(imread([pathname filename]))/255;



 fullpathname = strcat(pathname,filename);


 figure,imshow(Input_Image), title('Input Image')



 pause (2)
 %%
 RGB_G = rgb2gray(Input_Image);


 figure, imshow(RGB_G);


 G_NSE = imnoise(RGB_G,'gaussian',0,0.025);


 figure,imshow(G_NSE)
 pause (2)


 %%


 %after removing noise by Gaussian method



 FLTD_IMG = wiener2(G_NSE,[5,5]);


 figure, imshow(FLTD_IMG)
 pause (2)


 %%


 %using unsharpfiltering(The gray image is again converted to



 %using unsharp masking filter



 h=fspecial('unsharp');


 US_IMG=imfilter(Input_Image,h);


 figure,imshow(US_IMG),title('filtered iamge');
 pause (2)


 %% Boundary Extraction


 RGB_G(RGB_G>255)=0;


 S=strel('disk',4,0);%structuring element


 D=~im2bw(Input_Image);


 D=~im2bw(Input_Image);


 F=imerode(D,S);


 figure,imshow(D);title('binary image');
 pause (2)


 %%
 INP_SVMS = im2bw(Input_Image, graythresh(Input_Image));
 % imshow(bw)

 INP_SVMS = bwareaopen(INP_SVMS, 50);
 % imshow(bw)


 INP_GRP = bwlabel(INP_SVMS);
 INP_TR_ST = regionprops(INP_GRP,'PixelIdxList');


 % Initialize vector containing max values.

 MAX_VAL = zeros(numel(INP_TR_ST), 1);
 for k = 1:numel(INP_TR_ST)
 MAX_VAL(k) = max(Input_Image(INP_TR_ST(k).PixelIdxList));
 end


 INP_TR_S1 = find(MAX_VAL <150)



 INP_TR_ST=bwareaopen(INP_TR_S1,20)
 INP_TR_S1
 %%


 % se = strel('disk',1);
 % a_bw = imclose(INP_TR_ST,se);
 % a_bw=~INP_TR_ST;
 img2=im2bw(Input_Image,graythresh(Input_Image));
 figure, imshow(img2)
 pause (2)


 B = bwboundaries(img2);
 figure, imshow(img2)
 cc = bwconncomp(img2,4);
 number =cc.NumObjects;
 number
 s =regionprops(img2, 'Area');
 N_O = numel(s);
 N_O
 % INP_TR_S1
 hold on


 %% Image Classification


 SRC=im2double(Input_Image);


 In1=SRC(:,:,1);
 IN2=SRC(:,:,1);
 IN3=SRC(:,:,1);


 INP_S_VL=(In1+IN2+IN3)/3;


 SBL_VL = fspecial('sobel');
 SBL_VL_INV = SBL_VL';
 SB_FLT = imfilter(double(INP_S_VL), SBL_VL, 'replicate');
 SB_RPLKT = imfilter(double(INP_S_VL), SBL_VL_INV, 'replicate');
 SR_V = sqrt(SB_RPLKT.^2 + SB_FLT.^2);



 INP_CNV_FLT = deconvwnr(INP_S_VL,SBL_VL);


 LN_VAL = watershed(INP_CNV_FLT);
 LBL_IMG = label2rgb(LN_VAL);
 figure,imshow(LBL_IMG);title('Imege Analysis')


 STRL_FUN = strel('disk',20);
 IN_BW_OPN = imopen(INP_S_VL, STRL_FUN);


 IS_VL = imerode(INP_S_VL, STRL_FUN);

 IR_VL = imreconstruct(IS_VL, INP_S_VL);


 EN_FUN = imclose(IN_BW_OPN, STRL_FUN);



 IMG_DLT = imdilate(IR_VL, STRL_FUN);

 IMG_RCNST = imreconstruct(imcomplement(IMG_DLT),
imcomplement(IR_VL));
 IMG_RCNST = imcomplement(IMG_RCNST);
 figure, imshow(IMG_RCNST);title('Imege Analysis')


 IMG_RGNL_ILM = imregionalmin(IMG_RCNST);
 figure,imshow(IMG_RGNL_ILM);title('Imege Analysis')


 FLM_IMG = INP_S_VL;
 FLM_IMG(IMG_RGNL_ILM) = 255;
 figure, imshow(FLM_IMG);title('Imege Analysis')


 IMG_STRL_FUN = strel(ones(7,7));
 IM_FGM2 = imclose(IMG_RGNL_ILM, IMG_STRL_FUN);
 IMG_FGM3 = imerode(IM_FGM2, IMG_STRL_FUN);
 IMG_FGM4 = bwareaopen(IMG_FGM3, 20);
 FNL = INP_S_VL;
 FNL(IMG_FGM4) = 255;
 FNL1=im2bw(FNL);


 a_citra_keabuan = rgb2gray(Input_Image);
 threshold = graythresh(a_citra_keabuan);
 a_bww = im2bw(a_citra_keabuan,threshold);
 a_bw = bwareaopen(a_bww,80);
 se = strel('disk',2);
 a_bw = imclose(a_bw,se);
 a_bw=~a_bww;
 [LBL,N_O]=bwlabel(a_bw,8);
 LBL
 N_O


 if (N_O > 200)



 uiwait(msgbox('Stage I:The Segmented Image is Normal stage'));



 else
 uiwait(msgbox('Stage II: The segmented Image is affected by Cancer'));


 end
 clc

CONCLUSION
In this project, a classic image processing algorithm is designed to segment the
lesion picture for shape analysis. After demonstrating the dataset and method
used in this project, performance analysis shows that the algorithm can have an
even better performance on extracting shapes than the labels do. Finally, some
improvement suggestions are proposed to have further optimization of the
algorithm.

REFERENCES
1. M. J. Eide, M. M. Asgari, S. W. Fletcher, A. C. Geller, A. C. Halpern, W. R.
Shaikh, L. Li, G. L. Alexander, A. Altschuler, S. W. Dusza, A. A.
Marghoob, E. A. Quigley, and M. A. Weinstock; Informed (Internet course
for Melanoma Early Detection) Group, “Effects on skills and practice from a
 web-based skin cancer course for primary care providers,” J. Am. Board
Fam. Med. 26(6), 648–657 (2013).
2. 2. American Cancer Society. Cancer Facts and Figs. (2014).
3. 3. N. R. Telfer, G. B. Colver, and C. A. Morton; British Association of
Dermatologists, “Guidelines for the management of basal cell carcinoma,”
Br. J. Dermatol. 159(1), 35–48 (2008).
4. 4. A. Kricker, B. Armstrong, V. Hansen, A. Watson, G. Singh-Khaira, C.
Lecathelinais, C. Goumas, and A. Girgis, “Basal cell carcinoma and
squamous cell carcinoma growth rates and determinants of size in
community patients,” J. Am. Acad. Dermatol. 70(3), 456–464 (2014).
5. 5. E. A. Gordon Spratt and J. A. Carucci, “Skin cancer in
immunosuppressed patients,” Facial Plast. Surg. 29(5), 402–410 (2013).
6. 6. S. Ogden and N. R. Telfer, “Skin cancer,” Medicine (Baltimore) 37(6),
305–308 (2009).
7. 7. K. Korotkov and R. Garcia, “Computerized analysis of pigmented skin
lesions: a review,” Artif. Intell. Med. 56(2), 69–90 (2012).
8. 8. A. O. Berg, D. Best; US Preventive Services Task Force, “Screening for
Skin Cancer: recommendations and rationale,” Am. J. Prev. Med. 20(3
Suppl), 44–46 (2001).
9. 9. M. M. Rahman and P. Bhattacharya, “An integrated and interactive
decision support system for automated melanoma recognition of
dermoscopic images,” Comput. Med. Imaging Graph. 34(6), 479–486
(2010).
10.10. A. G. Isasi, B. G. Zapirain, and A. M. Zorrilla, “Melanomas non-
invasive diagnosis application based on the ABCD rule and pattern
recognition image processing algorithms,” Comput. Biol. Med. 41(9), 742–
755 (2011).
11.11. P. G. Cavalcanti and J. Scharcanski, “Automated prescreening of
pigmented skin lesions using standard cameras,” Comput. Med. Imaging
Graph. 35(6), 481–491 (2011).
12. J. A. Jaleel, S. Salim, and R. B. Aswin, “Artificial neural network based
detection of skin cancer,” IJAREEIE 1, 200–205 (2012). 13. M. Sadeghi,
M. Razmara, T. K. Lee, and M. S. Atkins, “A novel method for detection of
pigment network in dermoscopic images using graphs,” Comput. Med.
Imaging Graph. 35(2), 137–143 (2011). 14. C. Barata, J. S. Marques, and J.
Rozeira, “A system for the detection of pigment network in dermoscopy
images using directional filters,” IEEE Trans. Biomed. Eng. 59(10), 2744–
2754 (2012)