Scholarlink

Accurate segmentation and classification of brain tumors play a crucial role in diagnosis and treatment. A Self-Sparse Fuzzy Clustering (SS-FC) approach is introduced to improve brain tumor segmentation. For classification, an Attention-Based Ensemble Convolutional Classification (AECC) model integrates ResNet, Inception, and MobileNet as classifier variants. Comparative analysis shows that the SS-FC + AECC model achieves the highest accuracy of (97.8652%), outperforming Res + Inc. + Mob (95.6247%) and MobileNet alone (91.5674%). The SS-FC + AECC model also demonstrates exceptional sensitivity (97.9254%) and specificity (97.3258%), ensuring robust performance. With a precision of 96.8547% and an F1-score of 96.8754%, the model significantly outperforms other methods. Furthermore, it achieves a lower False Discovery Rate (FDR) of 3.1254% and a higher Negative Predictive Value (NPV) of 97.3258%. These results highlight the SS-FC + AECC model’s efficiency and reliability in brain tumor segmentation and classification, paving the way for improved clinical outcomes.

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

Attention-Driven U-Net Models for Accurate Brain Tumor Segmentation

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Baireddy Sreenivasa Reddy,
  • Anchula Sathish

摘要

Accurate segmentation and classification of brain tumors play a crucial role in diagnosis and treatment. A Self-Sparse Fuzzy Clustering (SS-FC) approach is introduced to improve brain tumor segmentation. For classification, an Attention-Based Ensemble Convolutional Classification (AECC) model integrates ResNet, Inception, and MobileNet as classifier variants. Comparative analysis shows that the SS-FC + AECC model achieves the highest accuracy of (97.8652%), outperforming Res + Inc. + Mob (95.6247%) and MobileNet alone (91.5674%). The SS-FC + AECC model also demonstrates exceptional sensitivity (97.9254%) and specificity (97.3258%), ensuring robust performance. With a precision of 96.8547% and an F1-score of 96.8754%, the model significantly outperforms other methods. Furthermore, it achieves a lower False Discovery Rate (FDR) of 3.1254% and a higher Negative Predictive Value (NPV) of 97.3258%. These results highlight the SS-FC + AECC model’s efficiency and reliability in brain tumor segmentation and classification, paving the way for improved clinical outcomes.