Scholarlink

<p>Brain tumor, a severe complicated disease caused by abnormal cell growth, necessitates a precise and timely diagnosis. This study reveals an advanced deep learning-based system for both classification and segmentation of brain tumor using Magnetic Resonance Imaging (MRI). A Convolutional Neural Network (CNN) is employed for tumor classification, which effectively distinguishing between different tumor types, while a U-Net model is utilized for precise tumor segmentation. To enhance segmentation accuracy, it is essential to obtain hyperparameter tuning; improving model convergence and precision for which Honey Badger Optimization (HBO) technique is employed. Additionally, data augmentation techniques are applied to address class imbalance, leading to improved classification performance. Key improvements include optimized parameter selection via HBO, which enhances segmentation accuracy compared to standard tuning methods, and robust CNN-based classification, ensuring reliable tumor differentiation. The application of data augmentation significantly improves model robustness, particularly in imbalanced data scenarios. Experimental results demonstrate an overall accuracy of 91.45%, with dataset-specific accuracy ranging from 87.10% to 99.54%, showcasing the effectiveness of these enhancements. The proposed system streamlines radiological workflows, reduces diagnostic workload, and improves early detection accuracy. Comprehensive validation ensures model reliability, safety, and compliance with medical standards. These advancements highlight the potential of AI-driven methods in medical imaging, demonstrating their impact on diagnosis of brain tumor.</p>

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

Distinguished approach for brain tumor segmentation and classification using U-net and customized convolution neural network in magnetic resonance images

  • 期刊论文

  • 发表日期：

    2026年5月27日

查看全文
  • S. Karkuzhali,
  • S. Bhuvan Raaj,
  • B. Hariharakrishnan

摘要

Brain tumor, a severe complicated disease caused by abnormal cell growth, necessitates a precise and timely diagnosis. This study reveals an advanced deep learning-based system for both classification and segmentation of brain tumor using Magnetic Resonance Imaging (MRI). A Convolutional Neural Network (CNN) is employed for tumor classification, which effectively distinguishing between different tumor types, while a U-Net model is utilized for precise tumor segmentation. To enhance segmentation accuracy, it is essential to obtain hyperparameter tuning; improving model convergence and precision for which Honey Badger Optimization (HBO) technique is employed. Additionally, data augmentation techniques are applied to address class imbalance, leading to improved classification performance. Key improvements include optimized parameter selection via HBO, which enhances segmentation accuracy compared to standard tuning methods, and robust CNN-based classification, ensuring reliable tumor differentiation. The application of data augmentation significantly improves model robustness, particularly in imbalanced data scenarios. Experimental results demonstrate an overall accuracy of 91.45%, with dataset-specific accuracy ranging from 87.10% to 99.54%, showcasing the effectiveness of these enhancements. The proposed system streamlines radiological workflows, reduces diagnostic workload, and improves early detection accuracy. Comprehensive validation ensures model reliability, safety, and compliance with medical standards. These advancements highlight the potential of AI-driven methods in medical imaging, demonstrating their impact on diagnosis of brain tumor.