A comprehensive review on machine learning in brain tumor classification: taxonomy, challenges, and future trends

In recent years, Machine Learning (ML), a key component of artificial intelligence (AI), has become increasingly popular in data analysis and processing. ML is now widely used in healthcare to detect and diagnose early-stage diseases, including brain tumors. This technology significantly improves brain tumor diagnosis and treatment accuracy and speed. However, it's important to note that the results from ML still require validation and approval from physicians and healthcare experts. The diverse techniques within ML and the critical nature of diagnosing tumor types and determining treatment regimens necessitate a thorough evaluation of the accuracy and reliability of ML approaches. Early diagnosis and accurate classification are among the most challenging aspects of using ML technology. Despite the importance of issues related to diagnosis and treatment in the healthcare field, as far as we know, no systematic, comprehensive, and detailed review has been conducted of ML approaches in diagnosing and classifying brain tumors based on the World Health Organization (WHO) standards. In this article, we conducted a systematic literature review (SLR) about ML approaches to diagnose and classify brain tumors. We utilized the WHO classical classification to identify widely used and unique approaches in this field. In addition, we provide important topics and open topics. Taxonomy is presented in four main areas based on WHO classifications: Grades I, II, III, and IV. This paper discusses ML approaches for diagnosing and classifying brain tumors according to important factors such as evaluation criteria, techniques, and advantages and disadvantages.