AN ENHANCED BRAIN TUMOR SEGMENTATION AND CLASSIFICATION USING IMPROVED K-MEANS WITH SWARM-BASED FIREFLY ALGORITHM

The number of medical cases connected to brain tumors has increased significantly in recent years, making it the 10th most prevalent type of tumor affecting both children and adults. However, if tumor identified in early stage, it is one of the most treatable types of cancer. As a result, scientists and researchers have been attempting to create advanced tools and procedures for determining such kind of tumor with their exact stage condition. Magnetic Resonance Imaging (MRI) and Computer Tomography (CT) are two commonly used procedures for sectioning and evaluating anomalies in the form, size, or location of brain tissues, which can aid in the detection of malignancies. The objective of this paper is to design and apply an Enhanced Brain Tumor Segmentation and Classification (EBSTC) model from MRI using with K-means with Swarm-based Firefly Algorithm (SFA) as a heuristic algorithm and here, Convolutional Neural Network (CNN) is used as a machine learning approach. A novel fitness function of SFA is introduced, which enhances the segmentation accuracy during Region of Tumor segmentation from MRI data. Experimental evaluation is performed against Brain Tumor Segmentation (BraTS) dataset and shown the effectiveness of improved K-means over k-means in terms of segmentation accuracy. We achieved an improvement of 4.54% in segmentation accuracy and 99.55% classification accuracy using CNN.