Scholarlink

Classification of brain tumours is crucial for computer-aided diagnostics (CAD) in health assessments. In light of the extensive procedures involved, manually identifying brain tumors using magnetic resonance imaging (MRI) is frequently labor-intensive and difficult, with the possibility of errors in detection and classification. Healthcare has significantly benefited from current developments in Deep Learning (DL), which have greatly enhanced the automation of medical image processing and diagnoses. One subclass of DL techniques, Convolutional Neural Networks (CNNs), is particularly good at visual learning and image categorization. To categorize brain tumours into three groups: gliomas, meningiomas, and pituitary tumours, we introduced the CNN method. We assessed the algorithm’s performance using a benchmark dataset and contrasted it with pre-trained models already in use, including VGG16, VGG19, ResNet50, MobileNetV2, and InceptionV3. According to the experimental findings, our suggested model had a high classification accuracy of 98.5%, with 99% precision, recall, and f1-score. These outcomes suggest that our approach accurately classifies the most prevalent brain tumours. The algorithm is a valuable tool to help clinicians identify brain tumours quickly and accurately because of its excellent generalization ability and speed of execution.

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 from MRI Scans Using a Novel CNN Architecture and Optimization Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shaik Jaffar Hussain,
  • B. Rupa Devi,
  • V. Mahalakshmi,
  • Harikala,
  • S. Z. Parveen,
  • S. Athinarayanan

摘要

Classification of brain tumours is crucial for computer-aided diagnostics (CAD) in health assessments. In light of the extensive procedures involved, manually identifying brain tumors using magnetic resonance imaging (MRI) is frequently labor-intensive and difficult, with the possibility of errors in detection and classification. Healthcare has significantly benefited from current developments in Deep Learning (DL), which have greatly enhanced the automation of medical image processing and diagnoses. One subclass of DL techniques, Convolutional Neural Networks (CNNs), is particularly good at visual learning and image categorization. To categorize brain tumours into three groups: gliomas, meningiomas, and pituitary tumours, we introduced the CNN method. We assessed the algorithm’s performance using a benchmark dataset and contrasted it with pre-trained models already in use, including VGG16, VGG19, ResNet50, MobileNetV2, and InceptionV3. According to the experimental findings, our suggested model had a high classification accuracy of 98.5%, with 99% precision, recall, and f1-score. These outcomes suggest that our approach accurately classifies the most prevalent brain tumours. The algorithm is a valuable tool to help clinicians identify brain tumours quickly and accurately because of its excellent generalization ability and speed of execution.