Scholarlink

The aggressive nature and crucial function of the brain make brain tumors a significant health concern. Effective and efficient treatment planning and an increase in survival rates are contingent upon the early detection of brain tumors. In automated brain tumor diagnosis, traditional DL techniques, especially CNN-based models, show promise but they are computationally and time-consuming to train. Additionally, this sophisticated approach lacks transparency. This work uses the Vision Transformer (ViT-B-16) model to present an enhanced methodology for brain tumor categorization. A total of 7,023 picture datasets were used to construct the model using several image processing methods. The performance of ViT-B-16 model was evaluated against those of VGG16, ResNet18 and AlexNet. The findings show that the ViT-B-16 model has an exceptional 99.92% diagnostic accuracy for brain tumors and offers useful information for the medical imaging sector and the medical decision-making process.

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

Vision Transformer for Brain Tumor Classification Using MRI Images: Performance and Interpretability Over CNN Models

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Badol Shikdar Rifat,
  • Faisal Imran,
  • Prity Ahoshan,
  • Abdus Salam,
  • Abu Khalid Siddique

摘要

The aggressive nature and crucial function of the brain make brain tumors a significant health concern. Effective and efficient treatment planning and an increase in survival rates are contingent upon the early detection of brain tumors. In automated brain tumor diagnosis, traditional DL techniques, especially CNN-based models, show promise but they are computationally and time-consuming to train. Additionally, this sophisticated approach lacks transparency. This work uses the Vision Transformer (ViT-B-16) model to present an enhanced methodology for brain tumor categorization. A total of 7,023 picture datasets were used to construct the model using several image processing methods. The performance of ViT-B-16 model was evaluated against those of VGG16, ResNet18 and AlexNet. The findings show that the ViT-B-16 model has an exceptional 99.92% diagnostic accuracy for brain tumors and offers useful information for the medical imaging sector and the medical decision-making process.