Scholarlink

In neuro-oncology, categorization of brain tumors is essential for prompt diagnosis and efficient treatment planning. In order to improve brain tumor classification, this work presents an innovative technique that blends Generative Adversarial Networks (GAN) with prominent Deep Learning models, such as Convolutional Neural Networks (CNN), VGG16, and ResNet50. Through the resolution of class imbalance and enhancement of model generalization, this approach guarantees a more precise classification of pituitary tumors, gliomas, and meningiomas using magnetic resonance imaging. Preprocessing methods and transfer learning make use of pre-trained networks for effective feature extraction, which speeds up training and improves performance. The effectiveness of the model is assessed using critical performance measures such Peak Signal-to-Noise Ratio (PSNR), sensitivity, specificity, and accuracy. Achieving an accuracy of 99.52% for specific tumor types, the combination of GANs with CNN-based architectures greatly enhances classification performance. The model’s ability to manage noise in images and improve picture quality during classification is further confirmed by PSNR analysis. This work shows how GAN-enhanced deep learning models used to give automated brain tumor categorization that is precise, scalable, and interpretable a benefit that will likely lead to better neuro-oncology therapeutic results.

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 GAN with Deep Learning Models

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sudha S. Tuppad,
  • Vidya S. Handur,
  • Vishwanath P. Baligar

摘要

In neuro-oncology, categorization of brain tumors is essential for prompt diagnosis and efficient treatment planning. In order to improve brain tumor classification, this work presents an innovative technique that blends Generative Adversarial Networks (GAN) with prominent Deep Learning models, such as Convolutional Neural Networks (CNN), VGG16, and ResNet50. Through the resolution of class imbalance and enhancement of model generalization, this approach guarantees a more precise classification of pituitary tumors, gliomas, and meningiomas using magnetic resonance imaging. Preprocessing methods and transfer learning make use of pre-trained networks for effective feature extraction, which speeds up training and improves performance. The effectiveness of the model is assessed using critical performance measures such Peak Signal-to-Noise Ratio (PSNR), sensitivity, specificity, and accuracy. Achieving an accuracy of 99.52% for specific tumor types, the combination of GANs with CNN-based architectures greatly enhances classification performance. The model’s ability to manage noise in images and improve picture quality during classification is further confirmed by PSNR analysis. This work shows how GAN-enhanced deep learning models used to give automated brain tumor categorization that is precise, scalable, and interpretable a benefit that will likely lead to better neuro-oncology therapeutic results.