This review explores the advantages of transformers and their application in medical image segmentation. Convolutional neural networks (CNNs) have been widely used in medical image segmentation, with U-Net being one of the most popular models. However, CNNs have limitations in capturing long-distance dependencies, which are crucial for medical image analysis. Transformers, originally developed for natural language processing, can capture long-distance dependencies and have been successfully applied to image classification tasks. Recent studies have extended transformers to medical image segmentation, leading to improved performance. Transformers use self-attention mechanisms to process global context information, making them effective for capturing long-range dependencies. Vision Transformers (ViT) have been applied to medical image segmentation, demonstrating significant improvements over traditional CNN-based models. Swin Transformers, which use local attention mechanisms, have also been applied to medical image segmentation, achieving state-of-the-art results. Several transformer-based models have been developed for medical image segmentation, including TransUNet, UNETR, and Swin-UNETR. These models leverage the strengths of transformers to capture global context and improve segmentation accuracy. However, transformers require large datasets and have challenges in capturing local features, which are essential for medical image segmentation. The review also discusses evaluation metrics for medical image segmentation, including pixel accuracy, mean pixel accuracy, Jaccard index, and Dice coefficient. These metrics are used to assess the performance of segmentation models. In conclusion, transformers have shown great potential in medical image segmentation, but challenges remain in capturing local features and handling small datasets. Future research should focus on optimizing transformer models to better balance global and local information, improving their performance in medical image segmentation tasks.This review explores the advantages of transformers and their application in medical image segmentation. Convolutional neural networks (CNNs) have been widely used in medical image segmentation, with U-Net being one of the most popular models. However, CNNs have limitations in capturing long-distance dependencies, which are crucial for medical image analysis. Transformers, originally developed for natural language processing, can capture long-distance dependencies and have been successfully applied to image classification tasks. Recent studies have extended transformers to medical image segmentation, leading to improved performance. Transformers use self-attention mechanisms to process global context information, making them effective for capturing long-range dependencies. Vision Transformers (ViT) have been applied to medical image segmentation, demonstrating significant improvements over traditional CNN-based models. Swin Transformers, which use local attention mechanisms, have also been applied to medical image segmentation, achieving state-of-the-art results. Several transformer-based models have been developed for medical image segmentation, including TransUNet, UNETR, and Swin-UNETR. These models leverage the strengths of transformers to capture global context and improve segmentation accuracy. However, transformers require large datasets and have challenges in capturing local features, which are essential for medical image segmentation. The review also discusses evaluation metrics for medical image segmentation, including pixel accuracy, mean pixel accuracy, Jaccard index, and Dice coefficient. These metrics are used to assess the performance of segmentation models. In conclusion, transformers have shown great potential in medical image segmentation, but challenges remain in capturing local features and handling small datasets. Future research should focus on optimizing transformer models to better balance global and local information, improving their performance in medical image segmentation tasks.