Recent Trends for Practicing Machine Learning in Brain Tumors: A Survey

Abstract

Brain tumors are considered to be a fatal disease that affects all age groups around the world. Abnormal formation of cells inside the brain results in brain tumors that may affect the normal functioning of the brain and hence requires urgent clinical attention. The early diagnosis of tumors may help in better clinical management and hence improve the survival rate of patients. Diagnosis reports based on biopsy procedu res are considered to be valid reports as per World Health Organization. But there were several studies reported in the literature that biopsy-based reports may contain errors which may be due to sampling or inter and inter-observer variability. Machine and deep learning-based MR methods in diagnosing brain tumors are researched nowadays as an alternative promising approach. This paper presents a survey based on machine and deep learning-based methods which tries to solve the problems of brain tumor diagnosis and classification. The paper also presents a comparative analysis of some of the approaches discussed in the literature. It was well established by the presented paper that although various machine and deep learning approaches were developed for diagnosing brain tumors, still need to be validated on global multicentric databases.