Enhanced Brain Tumor Image Segmentation Using U²-Net with Dual Attention Mechanisms

Abstract

Recent advancements in the field of deep learning have fundamentally transformed the landscape of medical image segmentation, particularly in the critical area of brain tumor diagnostics. This study introduces DAU²-Net, a novel dual attention-enhanced U²-Net architecture that integrates spatial and channel attention mechanisms with MobileViT blocks to prioritize tumor-specific features and model global contextual relationships. The primary objective of this integration is to significantly enhance the segmentation accuracy, thereby providing more reliable and precise diagnostic tools for medical professionals. Evaluated on BraTS 2019, DAU²-Net achieves a state-of-the-art Dice coefficient of 0.92. This remarkable result not only surpasses the performance of the traditional U-Net model, which recorded a Dice coefficient of 0.88 (+4%), but also outperforms the Fully Convolutional Network (FCN) with a Dice coefficient of 0.85 (+7%). And it also achieves 94.2% sensitivity and 97.5% specificity. These outstanding results highlight the significant effectiveness of the dual attention mechanisms utilized in DAU²-Net, which excel at capturing multi-scale contextual information. This capability is crucial for achieving precise tumor delineation, ultimately contributing to more accurate diagnoses and improved patient outcomes in the field of brain tumor diagnostics.