An automated brain tumor detection and classification from MRI images using machine learning techniques with IoT

In medical imaging applications, the accurate diagnosis of brain tumors from magnetic resonance imaging (MRI) at an early stage is a challenging task to researchers nowadays. The early detection of the tumor reduces the mortality rate from brain cancer-related deaths. Among various medical imaging techniques, the MRI is utilized due to low ionization and radiation, but manual inspection takes a lot of time. This proposed work introduces a machine learning technique (MLT) to recognize and classify the tumorous or non-tumorous regions based on the brain MRI dataset. There are four steps to carry out the MLT such as preprocessing, segmentation, feature extraction, and classification procedures. In the first stage, the skull is removed manually to reduce the time complexity by avoiding the process of the unwanted area of the brain image, and median filtering is utilized to filter the noise factor. Next, Chan-Vese (C-V) technique is used to segment the active tumor by selecting the exact initial point. In the very next step, the features of the tumor area are extracted using the gray level co-occurrence matrix (GLCM), and then important statistical features were chosen. Finally, a two-class classifier is implemented using the support vector machine (SVM) and its performance is then validated with k nearest neighbor (KNN). The accomplishment of the proposed flow work was evaluated in terms of accuracy, sensitivity, specificity, and precision by performing on the BRATS 2017 benchmark dataset. The simulation results reveal that the proposed system outer performs over existing methods with high accuracy.