Detection of tumor region in MR image through fusion of Dam construction and K-mean clustering algorithms

The Brain is the major central monitoring region of the human nervous system. Tumor growth is a speedy and abandoned growth of unhealthy cells of the brain. Magnetic Resonance Imaging (MRI) is the method frequently used to identify tumor region in the brain. Usually, MR images are not that much appropriate for comprehensive examination, because of this reason medical image treating is one of the most significant, demanding and rising field during this era of modern medical technology. The fundamental goal of this proposed technique is to highlight the borders of tumor area. This research technique is based upon three major tasks filtration, Edge preservation, and Segmentation. For this purpose first step is to remove `salt and pepper' noise from MR scan by applying median filter because of the efficiency, then preserve the edges in order to enhance the borders of tumor region. For this operation combination of two edge operators `Sobel and Canny' are used to design hybrid edge operator and display all the entities present in MR scan. Segmentation is one of the most important techniques required to evaluate and diagnose the brain tumor. Segmentation techniques are used to split an image in multiple regions. In segmentation pixels are labeled with same characteristics many different methods are used for segmentation of medical images just like C-means clustering, K-means clustering, region merging, region splitting and watershed segmentation etc. This proposed technique is the combination of three major segmentation approaches `K-means clustering, Watershed algorithm and Otsu binarization. This paper has displayed the contrast between all techniques at each step and also implemented the hybridized result by extracting tumor region in MR image. Results have compared and evaluated on the basis of different image parameters like MSE and PSNR, etc.