A brain tumor occurs when there is an atypical proliferation of cells in the brain, resulting in abnormal growth. The survival rate of patients with brain tumors is difficult to determine due to their infrequent occurrence and various forms. Magnetic Resonance Imaging (MRI) plays a crucial role in identifying tumor sites, but manual detection is time-consuming and prone to errors. Innovative breakthroughs in artificial intelligence, particularly in the realm of deep learning (DL), have paved the way for the creation of DL models that utilize MRI images for diagnosing brain tumors. In this paper, a three-step preprocessing approach is proposed to enhance the quality of MRI images, along with a Convolutional Neural Network (CNN) based on the EfficientNet-B0 model for accurate diagnosis of glioma, meningioma, pituitary tumors, and normal images. The model is designed to be computationally efficient, featuring a small number of convolutional and max-pooling layers, which allows for swift training iterations. The model achieved a 95.81% accuracy in detecting glioma, 97.54% accuracy in detecting meningioma, 96.89% accuracy in detecting pituitary tumors, and 97.14% accuracy in detecting normal images when tested on a dataset of 3394 MRI images.