Enhanced Brain Tumor Detection with Efficient Net: Leveraging Pre-trained Networks for Improved Classification Performance

For timely diagnosis and treatment planning, brain tumors must be accurately categorized from medical images. EfficientNet model, a recent deep learning model for its fast processing of image classification problems, serves as the foundation of the novel approach we introduce in this paper for categorizing brain tumors. By proportionally scaling depth, width, and resolution with compound scaling, EfficientNet achieves peak accuracy within the model at reducing computational complexity. We apply the EfficientNet model to categorizing various types of brain tumours in MRI scans and evaluate its performance on a publicly available dataset of images of pituitary, meningioma, and glioma tumours. In order to warrant correct early diagnosis and treatment planning, brain tumours need to be properly classified through medical imaging. Here, we utilize the EfficientNet model, one of the advanced deep learning architectures that is widely recognized for its high-performance results in image classification, to introduce a new approach of brain tumour classification. Compound scaling, where the depth, width, and resolution are scaled equally, is how EfficientNet designs model accuracy with reduced computational complexity.