Enhanced Identification of Malarial Infected Objects using Otsu Algorithm from Thin Smear Digital Images

International Journal of Latest Research in Science and Technology Vol.1,Issue 2 :Page No159-163,July-August(2012) http://www.mnkjournals.com/ijlrst.htm ISSN (Online):2278-5299 Enhanced Identification of Malarial Infected Objects using Otsu Algorithm from Thin Smear Digital Images 1 Amit Kumar, 2 Prof. Arun Choudhary, 3 Prof. P. U. Tembhare, 4 Prof. C. R. Pote 1 PG Student, Vishveshwarya Institute of Engineering and Technology, Ghaziabad, India, 2 Assistant Professor, Vishveshwarya Institute of Engineering and Technology, Ghaziabad, India, 3 Assistant Professor, Priyadarshini College of Engineering, Nagpur, India, 4 Associate Professor, Priyadarshini College of Engineering, Nagpur, India, 1 amit_pcea52@yahoo.com, 2 rushtoarun@rediffmail.com, 3 punesh_tembhare@yahoo.com, 4 scrpote@gmail.com Abstract Malaria is a serious global health problem, and rapid, accurate diagnosis is required to control the disease. An image processing algorithm to automate the diagnosis of malaria in blood images is developed in this project. Manual counting of parasitaemia is tedious and time consuming and need experts. In this paper, we are developing an image classification system is to positively identify malaria parasites present in thin blood smears. We are proposing an automatic technique for Malaria parasites detection from blood images by extracting red blood cells (RBCs) from blood image and classifying as normal or parasite infected. In this approach Otsu thresholding on green or blue channel of the RGB blood image can be applied for blood cell segmentation. Morphological processing is applied on segmented binary image for accurately detecting Parasite as well as blood cell. It is a two phase process, in first phase we are calculating total number of parasites and in second phase we are calculating total numbers of RBC cells, used to calculate percentage parasitaemia. Keywords RBC, Parasite, Thresholding, Morphological processing, RGB Color Model. I. INTRODUCTION Most of the diseases are caused due to the blood. Malaria is one of those diseases. For the detection of the malaria, generally the blood samples are analysed manually with the help of microscope. Many of the researchers have contributed in this area regarding the analysis of blood cell images by using the different image processing techniques. This paper presents the approach for the detection of the parasites and RBC count in the available blood cell image for the evaluation of malaria disease. Image segmentation technique is used for the analysis of the blood cell image. This paper is organized as follows: Section II deals with the discussion on the available literature related to the blood cell image analysis of various diseases. Proposed approach is described in the section III. Section IV presents the experimental results. Conclusion and future scope is discussed in section V followed by the references. II. RELATED WORK In literature, different approaches are reported for the microscopic medical image analysis from disease detection point of view. Here, in this section, we are summarizing few of the approaches related to the medical image analysis. In general the images are the microscopic images. Makkapati and Rao [1] explored the segmentation for HSV color space. A scheme presented in [1] is based on HSV color space that segments Red Blood Cells and parasites by detecting dominant hue range and by calculating optimal saturation thresholds. Methods those are less computation-intensive than existing approaches are presented to remove artifacts. The scheme is evaluated using images taken from Leishman-stained blood smears. Sensitivity of the scheme was found to be 83%. The method operates in HSV space and is dynamic in the sense that relevant thresholds are determined from the statistics of the given image rather than keeping them fixed for all images. Schemes determine optimal saturation thresholds to segment RBCs and chromatin dots that are robust with respect to the color variability encountered. The work in [1] illustrates the use of color image processing techniques. Raviraja and et al. [2] introduces a blood image processing for detecting and classifying malarial parasites in images of Giemsa stained blood slides, in order to evaluate the parasitamia of the blood. To detect the red blood cells that are infected by malarial parasites, statistical based approach is used. To separate automatically the parasites (trophozoites, ISSN 2278-5299 159

schizonts and gametocytes) from the rest of an infected blood image, color, shape and size information are used and later the image is compared with infected images after transformation of image by scaling, shaping to reconstruct the image. The images returned are statistically analysed and compare to generate a mathematical base. Also the evaluation of the size and shape of the nuclei of the parasite is also considered. Ruberto et al. [3] introduces morphological approach to cell image segmentation more accurate than the normal watershed based algorithm. The used non-flat disk-shape structuring element enhanced the roundness and compactness to improving the accuracy of normal watershed based algorithm whereas flat disk-shape structuring element to separate overlapping cells. These methods make use of knowledge of the RBC structure that is not used in existing watershed based algorithm. Sadeghian et al. [4] demonstrated a framework for segmenting white blood cells using digital image processing. This grey level image processing scheme has divided into two parts, first, nucleus segmentation based on morphological analysis, and then cytoplasm segmentation is based on pixel-intensity thresholding (Zack thresholding). In [5] a scheme based on RGB color space that segments Red Blood Cells and parasites by detecting dominant hue range and by calculating optimal saturation thresholds is presented. Methods that are less computationintensive than existing approaches are proposed to remove artifacts. The scheme is evaluated using images taken from Leishman-stained blood smears. Sensitivity of the scheme is found to be 83%. Automated image analysis-based software Malaria Count for parasitemia determination, i.e. for quantitative evaluation of the level of parasites in the blood, has been described in [6]. The presented system is based on the detection of edges representing cell and parasite boundaries. The described technique includes a pre-processing step, edge detection step, edge linking, clump splitting, and parasite detection. S.P.Premaratnea at el. [8] used digital images of oil immersion views from microscopic slides captured though a capture card. They were pre-processed by segmentation and greyscale conversion to reduce their dimensionality and later fed into a feed forward back propagation neural network (NN) for training it. Digital images were segmented to 64 pixels X 64 pixels images to be used as a training data set. The other reason for the segmentation was to make sure that the ANN s was kept to the smallest possible size in order to achieve easier training. Automated malaria detection by flow cytometry in combination with fluorescence staining was previously investigated [8], but background noise limited the detection to 2000 par/ml of blood. However, in comparison, THG images have much higher detection. The malaria parasite infections can be specifically detected in infected red blood cells by imaging THG emission from the hemozoin using infrared femtosecond pulsed laser excitation. Existing technology suggests that a flow cytometry device could be adapted using THG emission for automated stain-free diagnosis based on parasitamia counts, which would also lower the minimum parasitamia levels that are detectable. A method by Chen Pan et al. [10] is based on imageretrieval to classify cell image from high similarity image databases. RGB color histogram of cell and two intensity histograms corresponding to those local regions compose feature vector represents the cell image. Kernel principal component analysis (KPCA) is utilized to extract effective features from the feature vector. The weight coefficients of features are estimated automatically using relevance feedback strategy by linear support vector machine (SVM). Classification depends on the decision distance obtained by SVM and the nearest centre criterion. 90.5% classification accuracy of the method when combined with standardized sample preparation and image acquisition. This help to enable consulting physicians and experts engage in interactive diagnosis easily and to automatically search pathology image records to support reliable decision in detecting and discriminating. Also most of the researchers worked on mammogram images for the detection of MCCs. Some of them are discuss here. A research conducted by Lee and et al. [11] presented other automatic detection and classification of MCCs. A block region growing and k-means clustering-based thresholding is employed to extract the breast region. Then, a blanket method finds and locates the suspicious areas of possible MCCs clusters. The MCCs detection module is developed to automatically extract the MCCs from the ROIs. Among the image processing techniques that are involved in this module are gradient enhancement, contrast enhancement and Gaussian filters. The segmentation of MCCs from the background is done using entropy-based thresholding. Shape cognition which is based on neural network-like shape recognition systems is introduced as a classification technique of MCCs. The system in [10] achieved as high as 95% classification rate with 93% detection rate. III. PROPOSED APPROACH On the basis of review of above literature for the microscopic blood cell images, it is found that different image segmentation techniques are applied on color images, grey images and binary images. The different types of color models also used like RGB, HSV and in that a G, B and H; S components are used respectively for the identification of parasites and RBC by using image processing concept. But for the identification of correct number of parasite and RBC, proper segmentation technique is required. Most of the authors have applied combine approach of various segmentation schemes with morphological processing to improve the results. Being the reported work in literature is focused on the image segmentation, there is a scope to design some methodology or an approach for microscopic image segmentation. ISSN 2278-5299 160

A. Proposed Scheme As the computational power increases, the role of automatic visual inspection becomes more important. Image processing and artificial intelligence techniques can be introduced that may provide a valuable tool for improving the manual screening of specimens. An automated system is therefore needed to complete as much work as possible for the identification of malaria parasites. The following figure shows the proposed methodology which shows the processing on the thin smear image. 4. Use the Otsu algorithm method on histogram (which gives the threshold value) to get the binary image. 5. Do the hole filling using morphology and apply dilation or erosion with disk specification. 6. Removal of small area contours using some particular threshold (it can be measured in terms of the total pixels covered by the contour shape), here the value of this threshold is 2000 and 100 for lab data and available data respectively. 7. Finally count the total number of contour shapes which will provide the total count of the RBC cells. We are using RGB color space model in this methodology. We first separate out the RGB image into three different layers R, G, and B. By performing the processing on G or B layer we can segment the infected RBC cells out of all. In an image processing context, the histogram of an image normally refers to a histogram of the pixel intensity values. This histogram is a graph showing the number of pixels in an image at each different intensity value found in that image. For an 8-bit greyscale image there are 256 different possible intensities, and so the histogram will graphically display 256 numbers showing the distribution of pixels amongst those greyscale values. Fig. 1 Systematic flow of proposed work The presented scheme basically serves two purposes namely, parasite identification and RBC count. Hence, the presented scheme is divided into two phases. In first phase, chromatin dots are detected and in second phase, total numbers of RBC are counted. Procedure to identify the parasites is as follows: 1. Collection of the microscopic blood cell images in RGB color space or model. 2. Extraction of B component image from RGB color space image. 3. Prepare the histogram of the B component image. 4. Use the Otsu algorithm method on histogram which gives the threshold value. 5. Calculated threshold value is not used directly to get the binary image, but some offset value is added to this threshold. This offset value is identified through some experimentation, and later it is found that the desirable value of this offset is 0.25 which is used in this procedure to get the binary image. 6. Do the hole filling using morphology and apply dilation with disk specification. 7. All contour shapes are filled with pseudo colors. 8. Finally count the total number of contour shapes with different pseudo colors which will provide the total count of the RBC cells. Procedure to identify the RBC count is as follows: 1. Collection of the microscopic blood cell images in RGB color space or model. 2. Extraction of B component image from RGB color space image. 3. Prepare the histogram of the B component image. Histograms can also be taken of color images either individual histogram of red, green and blue channels can be taken, or a 3-D histogram can be produced, with the three axes representing the red, blue and green channels, and brightness at each point representing the pixel count. The exact output from the operation depends upon the implementation it may simply be a picture of the required histogram in a suitable image format, or it may be a data file of some sort representing the histogram statistics. The selection of thresholding required to analyses the histogram of the given images which give you the best selection of thresholding. But for the automatic selection of thresholding we are adopting Otsu algorithm given below. B. Otsu Algorithm In computer vision and image processing, Otsu's method is used to automatically perform histogram shapebased image thresholding, or, the reduction of a gray level image to a binary image. The algorithm assumes that the image to be threshold contains two classes of pixels (e.g. foreground and background) then calculates the optimum threshold separating those two classes so that their combined spread (intra-class variance) is minimal. Algorithm: 1. Compute histogram and probabilities of each intensity level. 2. Set up initial ù i (0) and ì i (0) 3. Step through all possible thresholds maximum intensity Update ù i and ì i Compute 4. Desired threshold corresponds to the maximum ISSN 2278-5299 161

Otsu's thresholding method involves iterating through all the possible threshold values and calculating a measure of spread for the pixel levels each side of the threshold, i.e. the pixels that either falls in foreground or background. The aim is to find the threshold value where the sum of foreground and background spreads is at its minimum. out with Matlab 7.3 and 2 GB RAM over these 8 images and finally the implementation results are summarized as results shown in the Table 1. The graphical comparison for manual results for Lab data and automated results are depicted in Figure 3 and Figure 4. C. Morphological Processing Morphology is a broad set of image processing operations that process images based on shapes. Morphological operations apply a structuring element to an input image, creating an output image of the same size. The most basic morphological operations are dilation and erosion. In a morphological operation, the value of each pixel in the output image is based on a comparison of the corresponding pixel in the input image with its neighbours. By choosing the size and shape of the neighbourhood, you can construct a morphological operation that is sensitive to specific shapes in the input image. After converting the image into binary form some of the images created holes. Fig. 3 Comparisons Parasite Count Manual Vs Automated IV. EXPERIMENTAL RESULTS The two types of malaria generally found in India one is vivax and other is falciparum but mostly the vivax is common in most of the people. The scheme is evaluated using images taken from Leishman-stained blood smears. The images for the experiment collected from the Druv Pathology Lab, Laxminagar, Nagpur, Maharashtra (India) by the Dr. Sheela Mundhada. The image database is as follows. The images having resolution of 1280 960 resolution which will increase the computational complexity. The given images infected by the malaria parasite having a type of vivax and faiciparum respectively. In total, the database contains 8 red blood samples infected by malaria parasites as shown in figure 2. IMAGE:01 IMAGE:02 IMAGE:03 IMAGE:04 IMAGE:05 IMAGE:06 IMAGE:07 IMAGE:08 Fig. 4 Comparisons RBC Count Manual Vs Automated TABLE I SUMMARISED RESULTS Parasites Parasite RBC Count RBC Count Count Image Count ( Proposed (Manual) ( Proposed (Manual) Approach ) Approach ) IMAGE 1 4 40 5 36 IMAGE 2 4 34 4 28 IMAGE 3 5 29 7 35 IMAGE 4 2 43 4 38 IMAGE 5 16 42 18 39 IMAGE 6 18 42 23 40 IMAGE 7 20 40 22 38 IMAGE 8 18 41 18 39 Fig. 2 Original Blood Sample Images from Druv Pathology Lab The manual data with respect to total numbers of parasites, and total number of RBC count are observed in the laboratory. In research approach, implementation is carried V. CONCLUSION Here in the given approach we have proposed the method for segmentation of infected cells in blood smear images with adoptive thresholding. The adaptive thresholding ISSN 2278-5299 162

can be performing with the help of Otsu algorithm which give better result as compare to averaging technique. After implementation of the proposed approach for the Lab sample images and for the available image database, it is found that the parasites count is near about matching with the manual count while in the RBCs count, some more difference is observed. Therefore, it become necessary to check the RBC count with the different plane image, for instance, B component only, in place of G component. From observation, it is clear that the proposed approach required less time in comparison with the manual evaluation. But for the available database image, the relevant data is not available so it becomes difficult for us to evaluate our approach. In future, we may go for the evaluation of proposed approach with G component only for the parasite and RBC count. Presently we have used the RGB color space images, in future; we would like to can evaluate the proposed approach with different color spaces. REFERENCES [1] V. V. Makkapati and R. M. Rao, Segmentation of malaria parasites in peripheral blood smear images, Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2009, pp. 1361-1364, Apr. 2009. [2] S. Raviraja, G. Bajpai and S. Sharma, Analysis of Detecting the Malarial Parasite Infected Blood Images Using Statistical Based Approach, IFMBE Proceedings, 3rd Kuala Lumpur International Conference on Biomedical Engineering 2006, vol. 15, part 12, pp. 502-505, 2007. [3] C. Di Ruberto, A. Dempster, S. Khan and B. Jarra, Segmentation of blood images using morphological operators, Proceedings of 15th International Conference on Pattern Recognition Barcelona, Spain, vol. 3, pp. 3401, 2000. [4] F. Sadeghian, Z. Seman and A. R. Ramli, A Framework for White Blood Cell Segmentation in Microscopic Blood Images Using Digital Image Processing, Biological Procedures Online, vol. 11, no. 1, pp. 196-206, Dec. 2009. [5] Vishnu V. Makkapati and Raghuveer M. Rao Segmentation of malaria parasites in peripheral blood smear images IEEE 2009. [6] S. Raviraja1, Gaurav Bajpai1 and Sharma S Analysis of Detecting the Malarial Parasite Infected Blood Images Using Statistical Based Approach, Proceedings 15, pp. 502-505, 2007 [7] Sio, W.S.S, et al: Malaria Count: An image analysis-based program for the accurate determination of parasitemia. Journal of Microbiological Methods. 2006. [8] S.P.Premaratnea, N. D. Karunaweerab, and S. Fernandoc, A Neural Network Architecture for Automated Recognition of Intracellular Malaria Parasites in Stained Blood Films, 2003. [9] C. J. Janse and P. H. Van Vianen, Flow cytometry in malaria detection, Methods Cell. Biol. 42 Pt. B: 295 318, 1994. [10] C. Pan, X. Yan and C. Zheng, Recognition of Blood and Bone Marrow Cells using Kernel-based Image Retrieval, IJCSNS International Journal of Computer Science and Network Security, vol.6 no.10, October 2006. [11] [S. K. Lee, C-S. Lo, C-M. Wang and P-C. Chung, A Computer- Aided Design Mammography Screening System for Detection and Classification of Micro calcifications, International Journal of Medical Informatics, vol. 60, pp. 29-57, 2000. ISSN 2278-5299 163