Diagnosis of Brain Tumor Using ANN with Spatial Fuzzy Clustering and Active Level Set Contour

Abstract

In medical science, image segmentation is a crucial and demanding factor. It has massive usage in the diagnosis of brain tumors. Magnetic resonance imaging (MRI) scans are used to detect the unnatural tissues of the brain efficiently which is always a powerful tool. The brain tumor identification and segmentation process are very steady and long delayed operation, because of the natural physical composition of human body. This paper deals with the classification of brain disease from MRI neuro-cell images. For the image segmentation and detection of brain tumor, a further combination of efficient method is proposed in this paper which uses inbuilt fuzzy C-means clustering which is further rectified by spatial clustering. Along with this segmentation, the proposed technique follows anisotropic diffusion filter and histogram of images for image enhancement. To meet the goal of separating tumor from normal brain cells, along with spatial fuzzy C-means, active level set contour is used for level-based image segmentation. The extraction of features from brain tumor is done by using DWT. With these features, the Artificial Neural Network classifier is trained. The accuracy of the proposed method is 96.7% on the patients 699 MRI data images. The result of this proposed method has been evaluated based on accuracy, sensitivity and specificity which is better than many existing methods.