Brain tumor image recognition method based on convolutional neural network

This study focuses on brain tumors, a highly fatal malignant lesion of the cranial cavity, and proposes a brain tumor detection and automatic classification framework based on the VGG16 pre-trained model. By employing transfer learning, we incorporated global average pooling layers and fully connected layers into the model, and froze the last layer to prevent overfitting and underfitting. We used the focal loss function to optimize model training, enabling it to learn from difficult samples and down-weight simple samples. In performance evaluation, we compared the improved VGG-GAP model with the original VGG16, VGG19, and InceptionV3, and the results showed significant improvements in all performance metrics.