Brain Tumor Detection From MRI Images Using Bag Of Features And Deep Neural Network

The anomalous and uncontrolled division of cells in the brain is the cause of a brain tumor. MRI (Magnetic Resonance Imaging) is a vital and tedious task for physically completed brain tumor segmentation. Accurate recognition of the area and size of the brain tumor plays a significant role in beneficial tumor determination and treatment. Therefore, recognition of the brain tumors must be accurate and fast. This article presents a methodology that can help to improve brain tumor detection by leveraging advanced machine learning approaches such as Support Vector Machine (SVM) and Deep Neural Network (DNN). Preprocessing is done by gray scale and dataset resizing in publicly available MRI datasets. The technique which is proposed based on a method called Bag of Features (BoF) for the selection of features is introduced to save learning time. First, the handcrafted features are extracted and transform the MRI images into a two-dimensional spatial image using Flexible Wavelet Transform (FWT), the first known method for the classification of brain tumors. Then AlexNet model is used to extract spatial features. A single combined feature vector table (CFV) is then obtained by combining four feature vectors. Feature selection is done using BoF by evaluating the features from the histogram. For classification, the selected features are sent to K-Nearest Neighbor (KNN) and SVM, and the highest precision is achieved by SVM classifier to identify brain tumors.