Tumor is a distorted bundle of tissue cells in which cells multiply endlessly, with no control over cell proliferation. Patients with early disease detection and treatment have a higher life quality and a longer expectancy of life. MRI imaging is useful in brain tumor investigation, diagnosis, and therapy planning. It assists doctors in determining the previous stages of a brain tumor. Detecting brain cancers with MRI scans is difficult due to the complicated structure of the brain. The study in this research article tackles the essential healthcare dilemma of effectively and timely diagnosing Brain Tumors. Deep Learning study used for classifying glioma tumor(glio_tu), meningioma tumor(menin_tu), no tumor(no_tu) state of the brain and pituitary tumor(pitu_tu). The research entails meticulous data preprocessing, proprietary CNN model architecture adaptation, and fine-tuning a collection of medical images. This study aims to enhance model performance through extensive validation and hyperparameter adjustment. Preliminary experimental outcomes demonstrate having custom Convolutional-Neural-Network (custom-CNN) model is very much accurately differentiate between distinct brain tumor states, with a 0.9850 validation and loss of 0.05. According to the research technique used, the combination of recent advances in deep learning combined with image processing and predictive modelling presents an intriguing way to identify solutions. These unique methods can then be used to assess accuracy and precision in the domain of brain tumors and other domains.