To builds the endurance pace of the brain tumor patients and to have an improved treatment system in restorative picture preparing, brain tumor segmentation is basic technique for finding. The early and right conclusion of brain tumours assumes a significant job. Magnetic Resonance Imaging (MRI) method is the most famous non-intrusive strategy; in nowadays imaging of organic structures by MRI is a typical exploring system. For malignant growth determination the brain tumours segmentation should be possible physically from MRI, which gives the poor degree of exactness and identification. The classification of variations from the norm isn't unsurprising and clear however it is a tedious errand for doctor. These days, the issue of programmed segmentation and examination of brain tumours are significant research region. Anyway the recognition of tumor is a difficult assignment since tumor has complex qualities in appearance and limits. Manual segmentation of brain tumor for disease conclusion, from enormous measure of MR pictures created in clinical daily schedule, is a troublesome and tedious errand. There is a requirement for programmed brain tumor picture segmentation. This paper does the audit of various writings of brain tumor segmentation. For segmentation, generally utilized clustering calculation like fluffy c-means, k-means a few specialists utilized convolution neural system approach and GPM. The motivation behind each segmentation calculation is to accomplish precise and proficient framework grew, so that to discover tumor in least time with most extreme exactness.