Paper Title
ACCURATE BRAIN TUMOR CLASSIFICATION BY USING MOBILENETV1

Abstract
Brain tumors are prevalent among children and the elderly, representing a serious type of cancer characterized by uncontrolled growth of brain cells within the skull. Due to the heterogeneity of tumor cells, classifying them accurately is particularly challenging. Convolutional Neural Networks (CNNs) are widely employed for visual learning and brain tumor identification. This study introduces a CNN-based model using a dense variant of EfficientNet, combined with min-max normalization, to classify 3,260 T1-weighted contrast-enhanced brain MRI images into four categories: glioma, meningioma, pituitary tumor, and no tumor. The proposed network is an enhanced version of EfficientNet with added dense and dropout layers. Additionally, data augmentation paired with min-max normalization was applied to enhance the contrast of tumor cells. The dense CNN model’s advantage lies in its ability to accurately classify a relatively small dataset of images. Consequently, the proposed model demonstrates exceptional performance, achieving 99.97% accuracy during training and 98.78% accuracy during testing. With its high accuracy and favorable F1 score, this newly designed EfficientNet CNN architecture proves to be a valuable tool for brain tumor diagnostic decision-making. Keywords - Mobilenetv1, Brain Tumor, Confusion Matrix, Efficientnet, CNN, MRI