Structural Magnetic Resonance Image (MRI) is a useful technique to examine the internal structure of the brain. MRI is widely used for brain tumor detection as it gives a clear picture of brain soft tissues. Brain tumor identification and classification is critical and time consuming task, generally performed by radiologists. Brain tumor of different sizes and shapes can occur in any person. Extraction of exact tumor region and analysis of minute differences is difficult for humans. Digital image processing methodologies like preprocessing, segmentation and classification are useful to clinical experts for proper diagnosis of brain tumor types. This paper focuses on current trends in brain tumor detection using MRI images .Analysis of various state of the art machine learning and deep learning based methods is given. Available datasets and challenges are discussed. This extensive survey will be helpful for future research to develop better decision support system, beneficial to radiologists for accurate brain tumor diagnosis.