Scholarlink

Effective brain tumor classification is critical for efficient diagnosis and treatment planning in medical applications. This paper employs a Stacked Convolutional Neural Network (SCNN) model for brain tumor classification achieving cutting-edge performance. The suggested technique has been investigated using Br35H::Brain Tumor Detection 2020 dataset. The proposed model outperforms established architectures including MobileNetV2, DenseNet12, ResNet50, and VGG16 across all performance metrics. SCNN can have deeper feature extraction and improved learning capacity making them more powerful than simpler CNN architectures for complex image analysis. 99.47% accuracy, 99.65% precision, 99.30% recall, and a 99.48% F1-score are attained with the SCNN, which demonstrate superior capability in accurately classifying brain tumors with minimal false positives and false negatives. The outcomes emphasize the capability of the proposed model to significantly enhance clinical diagnostic processes.

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

Stacked CNN: Deep Networks for Brain Tumor Classification

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sreenubabu Dasari,
  • Janmenjoy Nayak,
  • Tripti Swarnkar

摘要

Effective brain tumor classification is critical for efficient diagnosis and treatment planning in medical applications. This paper employs a Stacked Convolutional Neural Network (SCNN) model for brain tumor classification achieving cutting-edge performance. The suggested technique has been investigated using Br35H::Brain Tumor Detection 2020 dataset. The proposed model outperforms established architectures including MobileNetV2, DenseNet12, ResNet50, and VGG16 across all performance metrics. SCNN can have deeper feature extraction and improved learning capacity making them more powerful than simpler CNN architectures for complex image analysis. 99.47% accuracy, 99.65% precision, 99.30% recall, and a 99.48% F1-score are attained with the SCNN, which demonstrate superior capability in accurately classifying brain tumors with minimal false positives and false negatives. The outcomes emphasize the capability of the proposed model to significantly enhance clinical diagnostic processes.