Brain tumor classification and detection using a hybrid deep learning model

Abstract

Dignosis of brain tumor is an important topic in medical area. It involves a combination of medical imaging techniques, clinical assessments and sometimes molecular analysis. However, classification of brain tumor can be accomplished by deep learning methods easily and accurately. Therefore, automatized medical systems integrated with deep learning are highly demanded nowadays. Deep learning is a subset of machine learning, which itself is a branch of artificial intelligence (AI). It's based on the idea of artificial neural networks which is inspired by the structure and function of the human brain. They are designed to learn and make decisions in a similar way to humans. Brain tumor classification with deep learning includes using neural networks to analyze medical images such as MRI scans and classify them into different categories based on the presence or type of tumor. In this study, a hybrid deep learning model was designed and examined by combining pretrained VGG16 model with Random Forest machine learning classification algorithm. While deep features were extracted with pretrained model, classification with these features was handled by machine learning algorithm. A brain tumor dataset was trained and tested with proposed hybrid deep learning model. At the end of the experiments, it was seen that hybrid model approach has given promising accuracies of 92.31% and 90.48% in training and validation parts respectively.

Keywords: Brain Tumor Classification, Brain Tumor Diagnosis, Convolutional Neural Network, Deep Learning, VGG16, Random Forest, Hybrid Deep Learning.