Brain tumour prediction is a vital and difficult task in the medical field. Hydrocephalus is a kind of tumour that can develop in the brain. Hydrocephalus symptoms include bleeding, trouble walking, poor memory, lack of reasoning, problems with slurred speech in public, impairment in the legs and urinary tract, and impairment in daily tasks. Hydrocephalus can come to anyone at any age, it most frequently affects young children and elderly people. This tumour is caused by the excess production in the brain’s cerebrospinal fluid and mostly affects people’s nervous system and the brain. Therefore, the Magnetic Resonance Imaging (MRI) is a medical technique that has been considered to identify the tumour. Despite the availability of state-of-the-art medical facilities for precise diagnosis and efficient medical care, there is still a lack of significant control over the death rate. Hence, we proposed a novel hybrid approach to the brain tumour detection, and classification. To determine the tumour is benign or malignant, the suggested system employed Fuzzy K-Means (FKM) clustering and convolutional neural networks (CNNs). Hybrid methodology when used on brain tumour real-time datasets, the clustering algorithm has proven to be highly accurate in identifying subsets of tumours. And in calculating the distance between each object across a vast array of datasets, the fuzzy-based membership function is utilized to display the minimum distance-based dataset partitioning. Convolutional neural networks (CNNs) are employed in object classification and to prevent multiple clusters from overlapping. The size of the tumour is then determined by predicting the tumour dataset. The classification accuracy 90% was attained by the suggested method, which is both better and higher than the old methods. Also accomplished with an error minimization strategy. The future work is finding the various types of tumours in the brain.