Scholarlink

Brain tumors have become a devastating illness, second only to cardiovascular disorders worldwide. Early identification and treatment of brain tumors save millions of lives worldwide. Current approaches to brain cancer evaluation often require human involvement, which is both time-consuming and subjective. Automated detection tools are therefore crucial for fast and accurate diagnosis. Many existing methods are obtrusive, underscoring the need for a fully automated deep learning system to classify brain tumors. In our research, we employ convolutional neural networks (CNNs) and five pre-trained models—ResNet-101, VGG-19, MobileNet, XceptionNet, and InceptionV3—on 7,800 MRI images, evaluated across different epochs. At going to higher epochs, the models experienced overfitting, resulting in saturated accuracy. Optimal accuracy was achieved at 50 epochs. To further enhance accuracy, we optimized the pre-trained models of ResNet-101, VGG-19, MobileNet, XceptionNet, and InceptionV3 using a bio-inspired algorithm, particle swarm optimization (PSO). Our study achieved an optimal accuracy of 98% and an F1-score of 98.2% with InceptionV3 after optimization through particle swarm optimization (PSO). These results demonstrate the efficacy of our approach in providing rapid and accurate brain tumor classification, addressing the pressing need for reliable diagnostic tools in clinical settings.

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

Brain Tumor Classification Using Enhanced CNN and Performance Optimization with PSO

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • K. V. Sridhar,
  • R. Mounica,
  • Vikas Kumar Tiwari

摘要

Brain tumors have become a devastating illness, second only to cardiovascular disorders worldwide. Early identification and treatment of brain tumors save millions of lives worldwide. Current approaches to brain cancer evaluation often require human involvement, which is both time-consuming and subjective. Automated detection tools are therefore crucial for fast and accurate diagnosis. Many existing methods are obtrusive, underscoring the need for a fully automated deep learning system to classify brain tumors. In our research, we employ convolutional neural networks (CNNs) and five pre-trained models—ResNet-101, VGG-19, MobileNet, XceptionNet, and InceptionV3—on 7,800 MRI images, evaluated across different epochs. At going to higher epochs, the models experienced overfitting, resulting in saturated accuracy. Optimal accuracy was achieved at 50 epochs. To further enhance accuracy, we optimized the pre-trained models of ResNet-101, VGG-19, MobileNet, XceptionNet, and InceptionV3 using a bio-inspired algorithm, particle swarm optimization (PSO). Our study achieved an optimal accuracy of 98% and an F1-score of 98.2% with InceptionV3 after optimization through particle swarm optimization (PSO). These results demonstrate the efficacy of our approach in providing rapid and accurate brain tumor classification, addressing the pressing need for reliable diagnostic tools in clinical settings.