Automated Brain Tumor Detection Model Using Modified Intrinsic Extrema Pattern based Machine Learning Classifier

In Medical Image analysis, brain tumor detection, segmentation, and classification of tumor region is most tedious and time-consuming task. MRI is an imaging technique to visualize anatomy structure of human brain which helps to find the tumor affected region for the researchers and clinical experts. In this paper, for the denoising of brain MRI improved CNLM (collaborative Non-Local Means) filter is used. For denoised image, the process of segmentation is performed by Modified Intrinsic Extrema pattern. By taking the segmented region, features are extracted by correlation-based HOG (Histogram of Gradient). The selected features are used for the image classification using Improved kernel based SVM classifier which includes linear, RBF, quadratic and polynomial kernels. Thus, this modelling of IKSVM is helpful in enhancing the accuracy of classification.