This comprehensive review undertakes a rigorous evaluation of contemporary methodologies deployed in brain MRI image classification, with a specific focus on tumour detection. Leveraging an exhaustive survey of the pertinent literature, we dissect and compare the effectiveness of a diverse spectrum of approaches, encompassing both established machine learning algorithms and emergent deep learning models. Our analysis meticulously dissects the efficacy of feature extraction and selection techniques, meticulously scrutinizing their impact on enhancing diagnostic accuracy within the domain of brain tumour classification. This exhaustive synthesis illuminates the current landscape of brain MRI classification, elucidating the strengths, limitations, and nascent frontiers for future exploration within this critical field.