TRANSFER LEARNING MODEL FOR BRAIN TUMOR DETECTION WITH SNAKE SEGMENTATION

As medical standards advance, the role of medical imaging in clinical diagnosis and monitoring becomes increasingly vital. This paper presents a novel approach for the detection and classification of brain tumors using MRI imaging, leveraging digital image processing and artificial intelligence with a focus on Transfer Learning. The methodology begins with pre-processing MRI scans using a Gaussian filter, followed by snake segmentation for image partitioning. The images are then de-noised using a Parallel Non-Local Mean filter. For classification, a hybrid model combining a pre-trained VGG16 network with a Convolutional Neural Network is proposed, capitalizing on the advantages of transfer learning. This model’s efficacy is benchmarked against conventional machine learning algorithms—SVM, KNN, RF—and other deep learning architectures, including CNN variations with AlexNet, self-attention, and additive attention. Implemented in Python on Google Colab, the model was trained and tested on a Kaggle dataset comprising 5,712 images. Metrics employed for evaluation included accuracy, precision, and recall, alongside execution time. The findings demonstrate that our proposed transfer learning-enhanced model achieved superior performance with an accuracy of 96.7%, precision of 96%, and recall of 95%, thus outstripping the other examined models in brain tumor detection efficiency.