AI-Enabled Early Detection of Brain Tumor Through AFF-U-Net

This study presents a new AI-based method for detecting and segmenting brain tumors in MRI scans using the AFF-U-Net deep learning model. By incorporating attention mechanisms, the model improves its ability to focus on tumor regions, leading to better segmentation results. It was trained on the BraTS 2020 dataset, which includes different MRI types like FLAIR, TL,T1CE, and T2, along with their corresponding labeled tumor areas, allowing a detailed study of tumor regions.The AFF-U-Net model uses Convolutional Neural Networks (CNNs) to effectively extract important features, employs max-pooling to reduce the image size, and upsampling to restore the image resolution. The attention gates in the model's decoder phase help the network focus on the most relevant areas by refining the connections between the layers. After training for five rounds using the Adam optimizer and a binary cross-entropy loss function, the model achieved an accuracy of 97.9<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> on test data.In Future, this research demonstrates the effectiveness of the AFF-U-Net architecture in detecting brain tumors, offering a useful tool for radiologists in early diagnosis. By improving segmentation accuracy, this model can potentially enhance patient outcomes.