Scholarlink

Brain tumor diagnosis through MRI scans is critical but traditionally relies on manual analysis, prone to delay and inconsistency. This paper introduces a multi-stage deep learning framework using an optimized Convolutional Neural Network (CNN) for automatic classification of brain tumors into glioma, meningioma, pituitary tumor, and no tumor classes. Our approach incorporates advanced data augmentation, architectural regularization, and early stopping strategies to address overfitting and generalization. Trained on a publicly available dataset, the model achieved a peak training accuracy of 99% and a test accuracy of 97.3%. Comparative analysis shows our model outperforms recent approaches in accuracy, precision, and robustness, demonstrating strong potential for clinical deployment.

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

Multi-Stage Brain Tumor Classification Using Deep Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Abhimanyu Dudeja,
  • Adithya Sankar,
  • P. Saranya

摘要

Brain tumor diagnosis through MRI scans is critical but traditionally relies on manual analysis, prone to delay and inconsistency. This paper introduces a multi-stage deep learning framework using an optimized Convolutional Neural Network (CNN) for automatic classification of brain tumors into glioma, meningioma, pituitary tumor, and no tumor classes. Our approach incorporates advanced data augmentation, architectural regularization, and early stopping strategies to address overfitting and generalization. Trained on a publicly available dataset, the model achieved a peak training accuracy of 99% and a test accuracy of 97.3%. Comparative analysis shows our model outperforms recent approaches in accuracy, precision, and robustness, demonstrating strong potential for clinical deployment.