The segmentation of the human brain magnetic resonance imaging MRI plays a highly decisive role in diagnosing numerous diseases like tumors, Alzheimer's disease, edema, dementia etc. But it is a very challenging task because of presence of noise in the MRI images and also because the boundaries between different tissues of the brain cannot be easily distinguished. Standard fuzzy c-means clustering FCM method is proposed to segment the brain MRI accurately and to handle the noise. There are many variants of FCM and one such variant is the Intuitionistic fuzzy c-means clustering algorithm IFCM. It incorporates the advantages of intuitionistic fuzzy set theory. The IFCM handles the uncertainty, but is not robust to noise as it does not consider any local spatial information. Hence, in this paper a novel approach, namely the Robust and improved intuitionistic fuzzy c-means clustering algorithm RIIFCM is proposed. This algorithm is robust to noise as it considers local spatial information. We have demonstrated the efficiency of the RIIFCM algorithm compared to six other algorithms used for the brain image segmentation. The segmentation is carried out on a simulated MRI brain image and we demonstrate that the RIIFCM algorithm outperforms the other existing algorithms by calculating the similarity indices, false positive ratio FPR and false negative ratio FNR. © 2018 IEEE.