Kidney tumors are either benign or malignant. Simple kidney cysts are the mostly common, a benign mass that is completely different from cancerous tissue. Renal cysts, which often occur as naturally, do not require treatment unless they are causing symptoms [1]. Normally, cells that are the basic building blocks of the body multiply in a controlled manner according to the needs of the body, whereas cancer is the uncontrolled and irregular proliferation of a cell type. Cancer spreads to tissues and organs and invades them [2]. The level of tumor growth in renal cancer is an important indicator of malignancy; renal abnormalities can be correlated with the homogeneity of tissue density for classification [3-6]. Size, growth rate, and heterogeneity or homogeneity of the tumor are important criteria for well targeted tumor treatment [7]. Therefore, the internal status, density, and homogeneity of the lesion are used as parameters in the diagnosis and classification of the lesion. Because manual measurements are time-consuming and exhibit high intra- and inter-operator variability, computer-aided imaging is needed [8,9]. There are many image processing studies on the diagnosis of renal cell cancer. However, segmentation of kidney tumors has rarely been addressed. Studies are mostly focused on segmentation from abdominal images [10-14]. A conventional user-guided, automatic image segmentation method was used to separate right and left kidneys from sagittal images. In that study, 83% success was recorded [15]. In another study, a deformation model represented by statistical information was used on gray scale images. Automatic and manual segmentation showed a similarity of about 86.9% [16]. A computer-assisted detection system was developed using abdominal computed tomography (CT) images of kidney segmentation. The abdominal CT images were digitized and the gray-level threshold method was used for renal segmentation, producing a sensitivity of kidney tumor detection of 85% [17]. A method of automatically performing kidney segmentation of two-dimensional kidney CT images was proposed. In this method, first, the kidney position determination was performed using density based Connected Component Labeling (CCL). This algorithm was applied to tomographic images of different sizes using the kidney and spinal position. Then, kidney segmentation was performed using a growth-based method [18]. A new automatic segmentation method that employs Expectation Maximization (EM) algorithm using tissue information based features is used to identify kidney tumors. The aim in the study was to make a strong distinction by grouping all the pixels between the tumor and the healthy kidney areas. The success of the study was assessed by the kappa factort and about 57% success was obtained [19]. In addition to these studies, there are also knowledge-based models related to adaptive region growing and deformation [20-23]. To extract kidney in abdominal CT scan have been proposed effective approach. Authors have been used template evaluation method and concept of intensity values of a pixel to separate the desired region from the original image [20]. Based on the model of region homogeneity have been developed on a region growing algorithm. The method was successfully tested on artificial images and on CT images [21].