An Improved EfficientNet-B0 Approach for Brain Tumor MRI Image Classification

For the task of brain tumor classification in MRI images, this paper proposes a classification model based on an improved EfficientNet-B0 network. Due to the complex pathological features and diversity of medical imaging data, higher requirements are placed on the feature extraction capability and generalization performance of classification models. To address this, we optimized the original EfficientNet-B0 network by introducing the MS-CAM attention mechanism and improving the MBConv module, enabling the model to capture multi-scale feature information more comprehensively, thereby enhancing classification performance and generalization ability. Experimental results show that the improved model achieved a classification accuracy of 99.31% on the validation set, outperforming the original EfficientNet-B0 and other lightweight models. This model can provide clinicians with an efficient and accurate auxiliary diagnostic tool, potentially advancing the development of automated brain tumor diagnosis.