Artificial Intelligence for Brain Tumor Classification Using MRI Imaging

Abstract

One of the most crucial disorders is that of brain tumor. In brain tumors, it is of utmost important to provide diagnosis early and accurately for an effective treatment and to increase the chances of survival. Magnetic Resonance Imaging (MRI) is very often used in brain tumor diagnosis, since it clearly depicts soft tissues in an elevated resolution image. Though manual interpretation is very tedious and is subjective leading to different diagnoses based on the human interpretation of image. This paper review study about the use of Artificial Intelligence (AI), primarily focused on machine learning and deep learning algorithms for classification of brain tumor using MRI imaging. This paper has analyzed most frequently used AI models namely Convolutional Neural Networks (CNNs), Support Vector Machines (SVM), Random Forests, andhybriddeep learning architectures for automated brain tumor detection and classification. Pre-processing techniques such as image normalization, segmentation, noise reduction and feature extraction that affect the performance of models are also considered. Review has looked upon publicly available datasets, evaluation metrics used and a comparative study which confirms that AI based algorithms are superior to traditional diagnosis methods. Although much advancement has been made, some limitations are still existing such as, lack of annotated data set, model generalization, interpretability problem, clinical integration issues etc. This review has summarized recent work in AI-based classification of brain tumors, so as to enable radiologists in getting fastest, most efficient and dependable diagnosis.