Optimizing Brain Tumor Mri Classification Using Modified Vgg16 Model

Background: Brain MRI is vital for diagnosing brain tumors, yet challenges like variations in tumor size, shape, noise, and artifacts complicate accurate classification. Existing models like ResNet50 and Inception-ResNetV2 often struggle with these complexities. Methods: To address these issues, a modified VGG16 model was developed. This custom model includes multiple convolutional layers designed to effectively extract complex features from MRI images, enhancing its ability to manage the inherent complexities of brain MRI data. Results: The modified VGG16 model outperformed existing models, achieving a high accuracy of 97.2% and a low loss of 0.2. These results indicate its superior performance in classifying brain tumor images, even in the presence of noise and variations. Conclusion: The modified VGG16 model provides a robust solution for brain tumor classification, offering improved accuracy and reliability, which can lead to better patient outcomes.