This notebook detects tumors in brain MRI scans using a convolutional neural network with data augmentation and transfer learning. This notebook has 3 models: the first one is your standard CNN, the second model added data agumentation, and the third model added the pre-trained model VGG16. The first model had an accuracy of 86%, the final model had an accuracy of 98%. This notebook is hosted on Kaggle and can be found here: https://www.kaggle.com/code/jarredpriester/detecting-brain-tumors-vgg16-accuracy-98
The purpose of this project is to learn how to apply a convolutional neural network to brain MRI scans and to learn how to improve a CNN model's performance.
I learned about data augmentation and pre-trained models. I learned how to start with a regular CNN model, add data augmentation, and add the pre-trained model.
The dataset has 139 images of MRI brain scans. The dataset was found on Kaggle.
archive(3).zip - dataset
detecting-brain-tumors-vgg16-accuracy-98.ipynb - Kaggle python notebook