Enhanced Frost Self Organizing Map Segmentation Based Gradient Boost Classification for Brain Tumor Detection

The infected area identification in brain tumor MRI images are significant task since the brain tumor is a serious disease, arises due to irregular increase of cells in brain. Early tumor detection makes the patients to recover life. A novel technique called Enhanced Frost Preprocessed Kohonen Self Organizing Map Segmentation based Intensified Gradient Boosting Classification (EFPKSOMS-IGBC) technique is introduced for accurate brain tumor detection with higher accuracy and lesser time consumption. The EFPKSOMS-IGBC technique consists of three steps namely preprocessing, segmentation and ensemble classification. In this technique, Enhanced Frost Filter removes the noisy artifacts in input MRI image and provides higher PSNR ratio. After preprocessing, the Kohonen Self Organizing Map Segmentation process segments the input preprocessed image for extracting features like texture, color, shape, and intensity. In the third step, an Intensified Gradient Boosting Classification is performed to categorize MRI images as normal or tumor based on extracted features. Intensified Gradient Boosting is an ensemble technique that uses the set of weak learners for classifying the MRI image into different groups. The outputs of weak classifiers are combined to attain strong classification outcomes for tumor detection with greater accuracy. Qualitative and quantitative results analysis show that the EFPKSOMS-IGBC technique provides superior performance in brain tumor detection.