Improved Brain Tumor Classification using InceptionV3 and EfficientNet-B2 on MRI Images

Abstract

A type of malignant growth that can manifest in the tissues that surround the brain is referred to as a “brain tumor”. It is possible to classify it as either cancerous or noncancerous. There are two types of brain tumors: primary tumors and secondary tumors. The former refers to tumors that originate within the brain, while the latter can spread to other parts of the body. The signs and symptoms of a brain tumor can change depending on its size, where it is located, and what kind it is.  Some of these include vision problems, hearing issues, and seizures. Different types of treatment methods are available for brain tumors, such as surgery, radiation therapy, targeted therapy, and chemotherapy. The patient's health and the grade and size of the tumor are some of the factors that are considered when choosing a course of action. The precise classification of brain tumors plays a vital role in the planning of effective treatments. Around the world, individuals are dying from these diseases. Recent developments in deep learning (DL) have led to the development of models that can accurately identify brain tumors using MRI scans. This study presents a method that uses two advanced DL models InceptionV3 and EfficientNet-B2 for the purpose of improving the classification of brain tumors. The proposed method performed better than the current techniques when compared to a public dataset. According to the results of the study, the two models EfficientNet-B2 and InceptionV3, were able to accurately classify brain tumors. The proposed method could be utilized to improve the accuracy of the diagnosis and planning of brain tumors. It can also be applied to other imaging classification tasks. The study demonstrates the application of DL methods used in the analysis of medical images. It shows the efficiency of the EfficientNet-B2 and the InceptionV3 in distinguishing brain tumors on MRI scans.