Extracting the Liver and Tumor from Abdominal CT Images

M.Jayanthi
PhD Student, Department of ECE. B.M.S college of Engineering, Bangalore.

Dr B.Kanmani
HOD/Department of TCE, B.M.S college of Engineering, Bangalore. detect and differentiate liver tumors. However, the examination is very expensive and therefore is not as popular as CT. Angiography is an invasive procedure with potential complications.

Abstract Medical imaging and computer aided diagnosis traditionally focus on organ or disease based applications. CAD is used by radiologists as a second opinion in detecting tumors, accessing the extent of diseases and making diagnostic decision. Automatic segmentation of tumor from CT images is difficult due to size, shape, position and presence of other objects with the same intensity present in the image. Therefore, first segment the liver so that tumor can then be segmented accurately from it. In this paper, an approach for segmentation of liver and tumor from CT Images is mainly used for computer aided diagnosis of liver is proposed. The method uses region growing with optimized threshold algorithm. The liver is segmented using region growing method that starts from a seed point automatically detected and efficiently close around the vessels and tumors. Keywords Computed tomography, preprocessing, region growing, Thresholding, liver segmentation.

I.INTRODUCTION The body is made up of many types of cells. Normally cells grow and divide to produce new cells in a controlled and orderly manner. As a result a mass of extra tissue called a tumor may develop. A tumor can be benign (not cancerous) or malignant (cancerous) .Cells in malignant tumors are abnormal [2]. The most common medical imaging studies for early detection and diagnosis of liver tumors include Ultra Sonography (US), Computed Tomography (CT), Magnetic Resonance (MR) Imaging and Angiography [1]. Although Ultrasonography is noninvasive and does not emit radiation, it is operator dependent. Furthermore, the images are nonspecific for diagnosing benign, malignant tumors and their subtypes based on echogenicity. MR Imaging with its endogenous high tissue contrast and multi planar capability can accurately

In CT images [1-2], tumors located in a liver are generally identified by intensity difference between tumor and liver. Due to advance in computer technology, image processing technology has been used to assist physicians in identifying cancer in liver. The intensity of the tumor can be lower or higher than that of the liver. The main problem of liver tumor detection is due to low contrast difference between tumor and liver intensity values. Several approaches have been used for segmentation of liver and lesion from CT image datasets. The intensity threshold-based liver segmentation methods [9] require few threshold values to be defined manually, the deformable model methods require large number of manual inputs, in statistical model-based methods, the search for optimal model requires too much time, Level set requires high computation, and Snake algorithm requires considerable processing time. Level set, Watershed and Active contour based segmentation requires points on the tumor to be defined manually. Most of these methods are time consuming, produce over segmentation and unsatisfied results and could not respond identically to different patients The goal of the approach presented in this paper is to extract liver and tumor in computer tomography images of the abdomen. The first step is the pre processing of the whole volume in order to correct the contrast and to increase the speed and accuracy of liver segmentation. The next step is the reconstruction of the liver volume in which we first extract a region of interest and apply the seeded region growing method . The liver region segmented by region growing algorithm is processed with morphological operations. In which the unmasked region is completely masked with structuring

element. The third step is to obtain the tumor using Thresholding algorithm. The rest of this paper is organized as follows: Section II describes the various terms relevant to the paper. Section III describes the proposed method in detail. Section IV presents the results and the discussion and finally Section V gives the conclusion. II. PRELIMINARIES A. CT Scan CT Scan [11] is a computerized tomography scanning method. It is an X-ray exam to produce special images of the body. The image retrieved from this exam mainly defines the shape, size and position of the internal body parts allowing for a diagnosis. Advantage of using CT scanning: 1. The quality of the image is improved. 2. Internal complex features can be precisely measured without destructive testing. 3. Parts are scanned in a Free State environment without applying stresses that could have damaged the delicate parts or give a warped display B Abdominal CT image Liver The liver is located in the upper right quadrant of the abdomen, right side of the stomach and above the intestines. It extends from the abdomen to the thorax. From the perspective of the observer the liver is located on the left side of the CT image. The liver is the largest organ in the abdomen and the entire human body. Liver segmentation based on CT image is a challenging task due to the presence of similar intensity objects in the abdomen with no clear delineation between these objects and the liver.

1.Right lung 2. right hepatic vein. 3. Liver. 4.left hepatic vein. 5 Stomach. 6.splenic flexure. 7. Rate. 8,.the left lung. 9,Aorta. A method for liver segmentation has been developed based on a combination of region growing and threshold algorithm. Before proceeding to segmentation process, the image contrast is enhanced. C. CT data set Our dataset is composed of 40 abdominal contrast enhanced CT images. They have been acquired at the Malathi Manipal Hospital in Bangalore with a siemen multi detector spiral CT. Images are stored into a PACS system in DICOM format. Our dataset contains patients with normal, fatty, cirrhotic, overextended livers and livers with cancer, so that we have to take into account a big anatomical and gray level variability in the data D. Pre-Processing The use of preprocessing is to eliminate the noise and preserve the edges. In order to increase the speed and accuracy of liver segmentation, abdominal CT images are to be preprocessed. In the pre-processing step, the contrast of the image is improved for well differentiation of the liver from its surrounding soft tissues with the similar intensity. In the proposed method, preprocessing is the combination of contrast enhancement and morphological smoothing. E. Region Growing It is also classified as a pixel-based image segmentation method [9] since it involves the selection of initial seed points. This approach starts with initial seed points and then examines neighboring pixels (using either 4-connectivity or 8connectivity) to determine whether the pixel neighbors should be added to the region. Usually region growing is used in an iterative manner at different locations inside the liver until the entire organ is segmented. F. Morphological Operations Morphology here is a tool for extracting image components with the help of structuring elements that are useful in the representation and description of region shape such as boundaries and skeleton. Most elementary morphological operations are dilation, erosion, opening and closing. In this paper, dilation

Fig .1 :Axial slice of CT image

and erosion techniques can be used. Erosion is a process used to shrink or thins the area of object in the image. Image details smaller than the structuring elements are filtered from the image. Dilation is a process used to grows or expand the area of object in the image. Image details larger than the structuring elements are filtered from the image. III. PROPOSED METHOD The proposed method of our work is explained in following lines. It consist of various steps and each will be explained in detail. 1.First, preprocess the original image. Without preprocessing, directly if we extract the liver leads to the following problem undesirable boundaries resulting from its adjacent organs are extracted. Here preprocessing nothing but image simplification and is the combination of contrast enhancement and morphological smoothing. 2. For the preprocessed image, investigate and analyze the intensity distribution. So histogram of the preprocessed image is drawn and analyzed. For the fast and accurate liver segmentation, consider a prior knowledge of the liver on abdominal CT image such as shape, location and intensity value. From the prior knowledge, we can conclude the following things (i)Liver is located on the left side of the CT image. (ii)Liver is the largest organ in the abdomen and occupies more area. (iii)Right bottom region of the image is discarded because that portion does not contain the liver. 3. Preprocessed image is the simplified image and it consist of several other organs with the intensity similar to that of intensity of liver. These several organs are removed by morphological erosion. 4. Since the tumor is located inside the liver area. It is very important to identify the liver prior to detecting the tumor. In this paper, pixel based image segmentation algorithm nothing but region growing algorithm is used to segment the liver. The region growing algorithm is described in as: (i) Selection of seed points: A Seed point is the starting point for region growing and hence its selection is very important for the final result. Most of the region methods, select the seed point manually in advance. In this proposed method, accurate seed

point selection is used. First, determine the centroid of the largest connected region of the eroded image. Next, coordinates of the centroid point acts as an initial seed point for region growing. (ii) Segmentation starts from the selected seed point. 4 connected neighborhoods of the seed point region are determined.

(iii) The similarity measure of each neighborhood pixel with the seed point region is calculated. The pixel with the minimum distance is added to grown region. (iv) The process continues iteratively by comparing all neighbor adjacent pixels of the grown region using the similarity measure. This process stops when the intensity difference between region mean and new pixel becomes larger than a certain threshold. 5. The liver region segmented by region growing algorithm is processed with morphological operations. In which the unmasked region is completely masked with structuring element. The completely masked liver region is multiplied with the original input image and liver region is obtained. 6. The threshold algorithm [7] takes user defined mask as input. The intensity of the tumor region is determined. The region within the intensity range of the tumor is obtained by threshold algorithm. The morphological operation with structuring element is used to completely mask the tumor region. The tumor region is multiplied with the input slice image to get the tumor region.

IV. RESULTS AND DISCUSSION The image data has been acquired using CT system. The images are provided in dicom format with 512512 pixel matrix. Our segmentation method of liver was tested with five abdomens of five patients. To evaluate our proposed method, in this section, experimental segmentation of liver slices were carried out by using MATLAB 9. The sample shows five images that are Original Image, Region growing output , Morphological Image, Liver Region and Tumor detection

2(a)

2(b)

2( c)

2(d)

[3] Fujimoto H., Gu L., and Kaneko T., Recognition of abdominal organs using 3D mathematical morphology, Trans. Inst. Electron. Inf. Commun. Eng. D-II, no. 5, pp. 843-850, May 2001. [4] P. Campadelli, E. Casiraghi, E. Esposito, Liver segmentation from ct scans: a survey and a new algorithm., To appear in Journal of Artificial Intelligence in Medicine. [5] S.-J. Lim, Y.-Y. Jeong, C. Lee, Y.-S. Ho, Automatic segmentation of the liver in ct images using the watershed algorithm based on morphological filtering., Proceeding of SPIE 5370 (2004) 16581666. [6] S.-J. Lim, Y.-Y. Jeong, Y.-S. Ho, Automatic liver segmentation for volume measurement in ct images,Journal of Visual Communication and Image Representation 17 (4) (2006, August) 860875. [7] K Geetha ,S.Poonguzhali: Diagnosis of liver tumor using 3D segmentation methodfor selective Internal radiation therapy,IJAEEE ,vol2, 115-119. [8] Bae, K., Giger, M., and Chen, C.: Automatic segmentation of liver structure in CT images.Med. Phys. 20 (1993) 7178 [9]S.S kumar, R.S Moni,J rajeesh :Autoamtic segmentation of liver and tumor for CAD of liver JAIT ,vol 2 63-69. [10] Zhang, X., Tajima, T., Kitagawa, T., Kanematsu, M., Zhou, X., Hara, T., Fujita, H.,Yokoyama, R., Kondo, H., and Hoshi, H.: Segmentation of liver region with tumorous tissues. Proc. of SPIE Medical Imaging 2007: Image Processing Vol.6512 (2007) 651235-1-651235-9 [11] Dipak Kumar Kole : Automatic detection and size measurement of Hepatic Lesion ,IJWBC,Vol 1,93-98.

2(e)
Fig 2: Segmented Liver Region. (a) Original Image, (b) Region growing output ,( c) Morphological Image, (d) Liver Region.(e)Tumor detection

V CONCLUSION A new method for extracting the liver region and tumor is presented in this paper. The proposed algorithm uses Thresholding based on analysis of intensity distribution and morphological erosion to simplify the image which helps to decrease the computation time and efforts by removing the regions of other structures and tissues. The liver is segmented using confidence connected region growing that efficiently closes around the vessels and tumors. VI. REFERENCES
[1] Gao, L., Heath, D., Kuszyk, B., Automatic liver segmentation technique for three- dimensional visualization of CT data. Radiology 201 (1996) 359364 [2] Bae, K., Giger, M., and Chen, C.: Automatic segmentation of liver structure in CT images. Med. Phys. 20 (1993) 7178