A Combined Approach for Multiclass Brain Tumor Detection and Classification

Abstract

Brain tumor is a threat to human lives and is constantly growing. Early detection could reduce/minimized life threats. Currently, researchers are employing various machine vision-based techniques for brain tumor detection. This study focuses on a combined approach incorporating machine learning and deep learning for brain tumor detection. The initial step of the research was feature extraction which was acquired via a convolution neural network (Alex Net) and subsequently classification which was achieved via an ensemble classifier. The developed method is a non-invasive, contactless machine -vision based system for early diagnosing/detection of brain tumor. Various statistical variables such as mean, median, mode, skewness, and kurtosis to develop a multiclass ensemble classification model.  The results exhibit that  proposed method is 95.547% efficient compared to other methods.