Classification of Magnetic Resonance Imaging (MRI) is an aggressive concept to handle mathematical relations like Fuzzy, Rough Soft sets evaluation from bio-medical brain images. Processing of Soft, Rough, and Fuzzy mathematical theories are of scientific importance in the field of Bio-medical applications. Some of the authors have introduced Rough, Soft, and Fuzzy sets with connected and mathematical relations associated with each other. In this paper, we propose an Improved Intuitionistic Cluster Ensemble Segmentation Approach (IICESA) for the identification of brain tumor based on weight of image pixel with associative noise reduction in image selection. This technique also reduces Magnetic Resonance (MR) of brain from selected region based on threshold, which is explored from pixel weight to produce segmentation of brain tumor tissue from bio-medical images. All selected regions created intutionistic fuzzy, rough sets to enable and predict brain tumor tissue from medical brain images. This results of the simulated proposed approach gives efficient accuracy when compared to traditional approaches like Fuzzy C-Means (FCM), Generalized FCM (GFCM), and Soft Rough based FCM (SRFCM), Intutionistic Rough FCM (IRFCM) based on analysis of results.