Abstract: Brain tumor segmentation plays a crucial role in the diagnosis and treatment planning of brain tumors. In recent years, deep learning techniques have shown remarkable success in medical image segmentation tasks. In this study, we propose the use of three deep learning architectures, namely ResNet, U-Net, and ResUNet, for brain tumor segmentation. ResNet is a popular deep residual network known for its ability to capture complex image features. U-Net is a widely used architecture for biomedical image segmentation, known for its effective encoding-decoding structure. ResUNet is a hybrid architecture that combines the advantages of ResNet and U-Net. We evaluate the performance of these architectures on a publicly available brain tumor segmentation dataset. The dataset consists of magnetic resonance imaging (MRI) scans of brain tumors, with annotated tumor regions. We preprocess the data and train the models using a combination of loss functions and optimization algorithms. We compare the segmentation results of ResNet, U-Net, and ResUNet in terms of accuracy, sensitivity, specificity, and Dice coefficient. The experimental results demonstrate the effectiveness of deep learning models in segmenting brain tumors. The ResNet architecture achieves high accuracy in capturing fine details and subtle tumor boundaries. The U-Net architecture effectively captures contextual information and produces accurate tumor segmentations. The ResUNet architecture combines the strengths of both ResNet and U-Net, achieving improved segmentation performance.
Keywords: Deep Learning, Brain Tumor Segmentation, ResNet, U-Net, ResUNet, Magnetic Resonance Imaging (MRI).
|
DOI:
10.17148/IJARCCE.2023.125249
