Leukemia Detection using Image Processing

Transcription

1 IJSRD - International Journal for Scientific Research & Development Vol. 5, Issue 01, 2017 ISSN (online): Leukemia Detection using Image Processing Mrs.V.Venmathi 1 Ms.K.N.Shobana 2 Ms.Akshaya Kumar 3 Mr.D.Gajesh Waran 4 2,3,4 UG Student 1,2,3,4 Department of Electronic & Communication Engineering 1,2,3,4 SNS College of Engineering, Coimbatore, India Abstract Leukemia is a type of cancer which causes death among human. Only its detection and diagnosis helps to increase its cure rate. Presently, identification of cancer cells or blood disorders is by inspecting the microscopic images visually. This is done by analyzing the variations in texture, geometry, colour and statistical analysis of images. This paper describes various feature extraction techniques that can be used to detect leukemia using microscopic blood sample images. Image analysis plays an important in this method. Here first the cell biology basics are discussed and then the implementation of our proposed technique is carried out. Since our aim is to provide the cheapest method, only images are used. The tool we have used for the detection of cancer cells is MATLAB. Key words: Leukemia, blood cells, edge detection, GLCM, Gabor, Wavelet, MATLAB WBC is protecting the body against the infectious diseases. The ratio between the WBC and RBC is 1:700 [10]. The WBC count is used to diagnose to infection in the human body. Therefore, when a human is affected by any infection, the WBC count will be increased rapidly. C. Platelets This is a cell fragment that is produced from the megakaryocytes i.e, the cell with large nucleus. These cells are in bone marrow. The normal count of platelets is 1,50,000-4,00,000 per micro liter (µl) or cubic millimeter (mm 3 ) [10]. There occurs undesirable bleeding when its count decreases below 20,000/µL. I. INTRODUCTION The undesirable growth and reduction in the normal blood composition leads to blood cancer. Blood cancer is of three kinds, namely Leukemia, Lymphoma and Myeloma [10]. A. Leukemia It is a cancer of the lymphoblast i.e. the white blood cells. It diagnosis the count of white blood cells. B. Lymphoma The lymph system is affected by this type of blood cancer. It is also a kind of white blood corpuscles. The replication of WBC leads to lymphoma. C. Myeloma It is a kind of cancer in the plasma cells. These plasma cells act as a immune to the infections. There cells are found in bone marrow of the mammals, it also produce antibodies. It stops the formation of normal blood cells by establishing the occurrence of group of damaged plasma cells over the bone marrow. II. BACKGROUND The blood cells comprises of RBC, WBC and PLATELET. Each cell has its own structure and specific functions. The blood cells mainly involves the function of carrying the oxygen, carbon dioxide, hormones, metabolic wastes and food particles. The blood cells does not allow any foreign particles to get through it. The WBC cells plays the role of abstracting the infections. A. Red Blood Cells These cells carries oxygen from the lungs to all parts of the body. B. White Blood Cells The white blood cells are large in size when compared to RBC but they are found few in numbers. The main role of D. Plasma Fig. 1: Blood cell and its composition The plasma is used to carry the materials that are required by the cells. It is usually straw- colour liquid. III. TRADITIONAL CANCER DIAGNOSED Diagnostic procedures for cancer may include laboratory tests. A. Different Types of Laboratory Tests The clinical chemistry deals with the chemical process which is used to measure the volume of the chemical components in the blood and the tissues. The components are blood glucose, electrolytes, enzymes, hormones, lipids, other metabolic substances and proteins [10],[4]. The blood and urine are the most commonly used specimens in clinical chemistry. There are different tests to diagnose the chemical components in the blood and urine. B. One of the more common laboratory test (blood tests) A variety of blood tests are used to check the blood substances level that indicate how healthy the body is and whether infection is present. For example, blood tests revealing elevated levels of waste products, such as creatinine shows that the kidneys are not working efficiently to filter those substances out [10],[2]. Other tests check the presence of electrolytes - chemical compounds such as sodium and potassium are important to the body's healthy All rights reserved by 804

2 functioning. Blood clots are quickly determined by coagulation. A complete blood count (CBC) measures the maturity level of the different blood cells in a given volume of blood. This is one of the most common tests performed. Red blood cells are important for carrying oxygen and fighting anemia and fatigue. White blood cells fight infection. White blood cell increase indicates the infection. Bleeding and bruising are prevented by platelets. IV. PROPOSED WORK From the literature, it is found that typical steps for the process of the proposed work are, Fig. 2: Overall steps of the proposed work A. Image Acquisition Blood image from slides will be obtained from nearby hospital with effective magnification. B. Preprocessing Noise occurs while image acquisition. The noise may be due to illumination or shadows that make region of interest (ROI) appear as blurred image region. During this process, image enhancement such as contrast enhancement will be done. C. Image segmentation Segmentation of white blood cell (WBC) and determine ROI that is nucleus for WBC only. So, focus will be on nucleus of WBC only. Determination the types of WBC should be done from the nucleus. Only lymphocytes and myelocytes should be considered for cancer diagnosis. Others are excluded. Once the blast cells are determined, then proceed to the next step. Images containing nucleus only will be taken for the analysis. This is to reduce errors since there are similar color scales in WBCs with other blood particles. V. FEATURE EXTRACTION The feature of the image is an important thing that helps for isolation of the common properties of the image and as well detecting and naming the regions. It is the primary characteristics of the image. Few are detected by visual appearance certain features are provided by the artificial methods. The luminance of a pixels and grey scale textual regions are the natural features. Image amplitude histograms and spatial frequency spectra are those examples for the artificial features. Isolation of common property in image and subsequent identification of few features image. In our work we use four different types of feature extraction techniques. They are Edge detection, GLCM, Gabor and Wavelet transform. A. Edge Detection The intensity of pixel in the contract is mainly found in the edges of the images. The amount of data can be reduced by the detection of the edges, also stores only useful information and required structural details are preserved. It consists of many steps as a procedure. The first step is the finding of edges by looking for local maxima of gradient of image. A gaussian filter that which is used to smoothes the images by reducing the noise and unwanted details and textures. A large numbers of edge detections are available, each are designed in such a way to be sensitive to certain kind of edges. Edges orientation, noise environment and edge detections are the edge detecting operations are the variables involved. There operators are optimized as horizontal, vertical and diagonal edges. Edge detection is difficult in error images but both noise and edge contain high frequency content. The operators are used in noisy images and so they are accurate localization of detected edges. Therefore they are of two types. 1) Gradient Based Edge Detection In this method, the edges of the images are detected by determining the maximum and minimum in the first derivative of images [9]. 2) Laplacian Based Edge Detection In this method, the edges are one- dimensioned in shape of a ramp. The location can be determined by calculating the detective of image. This method search for zero crossing in second derivative of image to find edges [9]. B. GLCM GLCM stands for Grey Level Co-occurrence Matrix. It is also known as Grey- Level Spatial Dependence Matrix. It is the method of statistical examining texture that describes the relationship between the spatial of pixel. This method involves characterizing the texture of image by calculation how often pairs of pixel with specific values measure from this matrix. The following statistics provides information about the texture of an image. 1) Contrast Measures the local variations in the gray-level cooccurrence matrix. Contrast= (i j) 2 i j g ij 2) Correlation Measures the joint probability occurrence of the specified pixel pairs. Correlation= i j (ij)g ij μ x μ y σ x σ y σ y are the means and standard where µ x, µ y, σ x and deviations of g x and g y 3) Energy Energy known as the angular second moment or uniformity gives the sum of squared elements in the GLCM. 2 Energy = i j g ij 4) Homogenity Measures the closeness of the distribution of elements in the GLCM to the GLCM diagonal. Homogenity= i j 1 1+(i j) 2 g ij All rights reserved by 805

3 5) Gabor Gabor filters are biologically motivated convolution kernels that have enjoyed widely usage in many applications in the field of image processing and computer vision e.g., face recognition, facial expression recognition, iris recognition, optical character recognition, vehicle detection etc. In order to extract local/ global spatial textural micro-patterns in ROIs, Gabor filters can be tune with different orientations and scales thus provide powerful statistics which could be very useful for blood cancer detection. The general function g(x,y) of 2D (for image) Gabor filter family can be represented as a Gaussian kernel modulated by an oriented complex sinusoidal wave. G(x,y)= 1. e [ 1 (x 2 2 σ2 +y2 x σ2 )] y. e (2πjwx ) 2πσ x σ y x = x. cosθ + y. sinθ and y = x. sinθ + y. sinθ Where σx and σy are the scaling parameters of the filter. The weighted summation describes the neighbourhood of the pixels. W is the central frequency of the complex sinusoidal and θ [0,π) is the orientation of the normal to the parallel stripes of the Gabor function. 6) Wavelet Transform The Haar wavelet, which Alfred Haar discovered in 1910, is both powerful and pedagogically simple. The basic Haar wavelet is a piecewise constant function. 1 0 t 1/2 φ(t) = { 1 1/2 t 1 0 otherwise Wavelet transforms provide a multi-resolution approach to texture analysis and classification. Recursive filtering and sub-sampling are involved in computation of the wavelet transforms. At each level, the signal is decomposed into four frequency sub-bands, LL, LH, HL, and HH, where L denotes low frequency and H denotes high frequency. LL LH HL HH Table 1: Level one of the 2D wavelet transform VI. RESULTS AND DISCUSSION In this section, the output of the feature extraction techniques obtained for the given input microscopic image are discussed. Fig 4: shows the input image containing cancer cells. Fig 5: shows the gray scale output of the input image. Fig 6: shows the output obtained from the Sobel edge detection techniques. Fig 7: shows the result of the GLCM technique used on the same input image. Fig 8: shows the output obtained using the Gabor filter bank. Fig 9: shows the output of the Haar wavelet transform. Fig. 5: Gray scale image Fig. 6: Sobel edge detected image Fig. 7: GLCM Fig. 8: Gabor magnitude Fig. 4: Input image Fig. 9: Gabor phase All rights reserved by 806

4 Fig. 10: Wavelet coefficients VII. CONCLUSION The primary goal of this work is to develop an automated system to detect blood cancer using Image processing. Blood cancer is the major causes of death globally and the early detection of this disease is very important. The computer aided cancer detection system helps the physician for cancer diagnosis. From the analysis it is examined that, sobel and prewitt edge detection operator works good for edge detection technique. GLCM produces the result based on four different criteria such as contrast, correlation, energy and homogeneity. In Gabor, magnitude and phase are calculated in order to extract the features of the image. Wavelet coefficients are calculated in the wavelet transform. It is concluded that these four feature extraction techniques are good in the leukemia detection. We hope that the proposed method will help physician or biologists for cancer diagnosis by reducing time and with more accuracy. VIII. FUTURE WORK In future, the extracted features from the existing output will be given to the SVM technique or the Fuzzy logic to produce the automated result. REFERENCES [1] Adollah.R, Mashor.M.Y, Nasir.N.F.M, Rosline.H, Mashin.H, Adillah.H, Blood Image Segmentation: A Review, Biomed 2008, Proceedings 21, 2008, pp [2] Deepika N. Patil and Uday P. Khot. Image processing Based Abnormal Blood Cells Detection, International Journal of Technical Research and Applications e- ISSN: [3] Fauziah Kasmin, Anton Satri Prabuwono, Azizi Abdullah, Detection of Leukemia in Human Blood Sample Based on Microscopic Images: A Study, Journal of Theoretical and Applied Information Technology, Volume 46 No.2, December [4] Himali P. Vaghela, Hardik Modi, Manoj Pandya, M.B. Potdar Leukemia Detection using Image processing Techniques, International Journal of Applied Information Systems ISSN: , Volume 10 No.1, November [5] Hirimutugoda.Y.M, Wijayarathna.G. Artificial Intelligence-Based Approach for Determination of Haematalogic Diseases, IEEE, [6] Malhar Bhatt, Shashi Prabha, Detection of Abnormal Blood Cells using Image Processing Technique, International Journal of Electrical and Electronics Engineers, ISSN (E), IJEEE, Volume 07, Issue 01, [7] Mohamed Ali, David Clausi Using The Canny Edge Detector for Feature Extraction and Enhancement of Remote Sensing Images, /01/$10.00 (C) 2001 IEEE. [8] Mohapatra.S, Patra.D, Satpathi.S, Image Analysis of Blood Microscopic Images for Leukemia Detection, International Conference on Industrial Electronics, Control and Robotics, IEEE, 2010, pp [9] Navin D. Jambhekar, Red Blood Cells Classification using Image Processing, Science Research Reporter 1(3): , Nov. 2011, ISSN: (online). [10] Nur Alom Taludkar, Daizy Deb, Sudipta Roy, Automated Blood Cancer Detection Using Image Processing Based on Fuzzy system, International Journal of advanced Research in Computer Science and Software Engineering, Volume 4, Isssue 8, August [11] Osowski.S, Siroic.R, Markiewicz.T, Siwek.K, Application of Support Vector Machine and Genetic Algorithm for Improved Blood Cell Recognition, IEEE Transactions on Instrumentation and Measurement, Vol. 58, No. 7, July 2009, pp [12] Piurri.V, Scotti.F, Morphological Classification of Blood Leukocytes by Microscope Images, IEEE International Conference on Computational Intelligence for Measurement Systems and Applications, Boston, MA, USA, July 2004, pp [13] Ritter.N, Cooper.J, Segmentation and Border Identification of cells in Images of Peripheral Blood Smear Slides, 30 th Australian Computer Science Conference, Conference in Research and Practice in Information Technology, Vol. 62, 2007, pp [14] Ruberto.C.D, Dempester.A, Khan.S, Jarra.B, Analysis of Infected Blood cell Images using Morphological Operators, Image and Vision Computing, IEEE Vol. 20, 2002, pp [15] S. Jagadeesh, Dr. E. Nagabhooshanam, Dr. S. Venkatachalam, Image Processing Based approach to Cancer Cell Prediction in Blood Samples, International Journal of Technology and Engineering Sciences, Vol.1(1), ISSN: [16] Sabino.D.M.U, Costa.L.D.F, Rizzatti.E.G, Zago.M.A, A Texture Approach to Leukocyte recognition, Real time Imaging, IEEE Vol. 10, 2004, pp [17] Shailesh J. Mishra, Mrs. A.P.Deshmukh. Detection of Leukemia Using MALAB, International Journal of Advanced Research in Electronics and Communication Engineering, Volume 4, Issue 2, February All rights reserved by 807

5 [18] Tek.F.B, Dempster.A.G, Kale.I, Parasite Detection and Identification for Automated Thin Blood Film Malaria Diagnosis, Computer Vision and Image Understanding, IEEE Vol. 114, 2010, pp [19] Valencio.C.R, Tronco.M.N, Domingos.A.C.B, C.R.B. Knowledge extraction using visualization of Heamoglobin parameters to identify thalassemia, proceedings of the 17 th IEEE symposium on Computer based medical systems, 2002, pp [20] Wongseree.W, Chaiyaratna.N, Thalassaemic Patient Classification Using a Neural Network and Genetic Programming, IEEE, 2003, pp All rights reserved by 808