Scholarlink

This study lays out a comprehensive analysis of the use of machine learning methods to detect and segment brain tumors in MRI scans. This study includes tumor identification using MRI scan by utilizing the ResNet50 model, and for precise tumor segmentation, it incorporates regional object recognition methodologies using U-Net. The goal is to accurately determine the specific kind of brain tumor, which must be classified into the three primary categories of Meningioma, Glioma, and Pituitary, this study attains a classification accuracy of at least 95%. This suggested model aims to enhance the identification process by employing a convolutional neural network with one hundred filters in the first layer, utilizing a dataset of 3,064 MRI scans accessible on the Figshare website (Cheng in Brain tumor dataset authored. Figshare Website). This method significantly reduces the burden on a surgeon, hence improving the precision of tumor identification and categorization beyond the specialization. The Resnet50 and U-Net methodologies effectively classify and segment tumors in MRI, as demonstrated by the results. A user-friendly graphical interface is available, facilitating the upload of a real MRI scan for tumor classification and segmentation. Nevertheless, the following limits remain: first, data accessibility; second, the generalizability of models across diverse instances; and third, the clinical validation of the used approaches.

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 Detection and Classification Using Machine Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ayesha Taranum,
  • N. Vedavathi,
  • N. Hemanth,
  • M. Pavan Kumar,
  • K. B. Praneeth,
  • M. Preetham

摘要

This study lays out a comprehensive analysis of the use of machine learning methods to detect and segment brain tumors in MRI scans. This study includes tumor identification using MRI scan by utilizing the ResNet50 model, and for precise tumor segmentation, it incorporates regional object recognition methodologies using U-Net. The goal is to accurately determine the specific kind of brain tumor, which must be classified into the three primary categories of Meningioma, Glioma, and Pituitary, this study attains a classification accuracy of at least 95%. This suggested model aims to enhance the identification process by employing a convolutional neural network with one hundred filters in the first layer, utilizing a dataset of 3,064 MRI scans accessible on the Figshare website (Cheng in Brain tumor dataset authored. Figshare Website). This method significantly reduces the burden on a surgeon, hence improving the precision of tumor identification and categorization beyond the specialization. The Resnet50 and U-Net methodologies effectively classify and segment tumors in MRI, as demonstrated by the results. A user-friendly graphical interface is available, facilitating the upload of a real MRI scan for tumor classification and segmentation. Nevertheless, the following limits remain: first, data accessibility; second, the generalizability of models across diverse instances; and third, the clinical validation of the used approaches.