Scholarlink

Brain tumor segmentation from MRI images forms a significant task in medical imaging. Many different methods have been proposed in the literature; however, this work compares three deep learning models, namely, U-Net, U-Net with ResNet, and U-Net++ , for automatic brain tumor segmentation on the MRI dataset. The baseline is U-Net, ResNet-enhanced U-Net guarantees the deeper learning of features, while U-Net++ incorporates dense skip connections to capture the details better. Our findings also provide evidence that U-Net ++ brings in the highest accuracy in segmentation since its complex architecture allows for huge enhancements in capturing detail. The enhanced precision by U-Net++ points out a new potential avenue for brain tumor segmentation and, in a broader sense, an improvement of accuracy and reliability in clinical outputs.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Segmenting Brain Tumors with Deep Learning: U-Net, U-Net with ResNet, and U-Net++

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jhanavi Malhotra,
  • Kavisha Gupta,
  • Kavya Singla,
  • Himanshu Mittal

摘要

Brain tumor segmentation from MRI images forms a significant task in medical imaging. Many different methods have been proposed in the literature; however, this work compares three deep learning models, namely, U-Net, U-Net with ResNet, and U-Net++ , for automatic brain tumor segmentation on the MRI dataset. The baseline is U-Net, ResNet-enhanced U-Net guarantees the deeper learning of features, while U-Net++ incorporates dense skip connections to capture the details better. Our findings also provide evidence that U-Net ++ brings in the highest accuracy in segmentation since its complex architecture allows for huge enhancements in capturing detail. The enhanced precision by U-Net++ points out a new potential avenue for brain tumor segmentation and, in a broader sense, an improvement of accuracy and reliability in clinical outputs.