A Systematic Review of Machine Learning and Deep Learning Techniques for Brain Tumor Classification, and Grading

Brain tumors rank among the most lethal forms of cancer, and their early detection is crucial for improving patient outcomes. Beyond timely identification, accurately determining tumor type and grade is essential for guiding treatment strategies. While Magnetic Resonance Imaging (MRI) is a mainstay in tumor detection and diagnosis, the precise classification of brain tumors remains challenging due to variations in their morphology, including shape and size. In recent years, artificial intelligence—particularly machine learning methods such as Support Vector Machines (SVM) and k-nearest neighbors (KNN), and deep learning architectures including Convolutional Neural Networks (CNNs)—has emerged as a powerful tool for classifying brain tumor types and grades from MR images. This review provides a comprehensive examination of contemporary studies that leverage these methods for tumor classification. The analysis is based on a structured search of major scientific databases and includes 81 studies published between 2014 and 2025 that met predefined inclusion criteria. Specifically, it analyzes the classification techniques employed, the datasets and their characteristics, feature extraction approaches, and performance evaluations reported in the literature. Reported results show that deep learning methods generally outperform traditional machine learning models, with many studies achieving accuracies above 90% in tumor type or grade classification tasks. In addition to summarizing existing work, this review provides a unified perspective by jointly examining tumor type and multi-class grade classification studies, highlighting understudied areas such as non-glioma grade prediction and common dataset limitations. By synthesizing existing findings, the review identifies key research gaps and proposes directions for future investigations. A key future direction highlighted in this review is the need for more comprehensive multi-class tumor grade prediction studies, particularly for non-glioma tumor types. The insights presented here aim to aid researchers and clinicians in developing more accurate and efficient brain tumor classification systems, leading to earlier diagnosis, reduced diagnostic variability, and more informed treatment planning in clinical practice.