Scholarlink

Brain Tumor (BT) is one of the most life-threatening diseases with a relatively low survival rate. Several diagnostic techniques, including magnetic resonance imaging (MRI), computed tomography (CT), biopsy, cerebral arteriogram, and positron emission tomography (PET), are used to identify BT. With the recent advancements in Deep Learning (DL), radiologists and pathologists can now detect and predict BT with greater accuracy. DL techniques are particularly well-suited for efficiently processing large volumes of image data from various scans. This research introduces a high-accuracy convolutional neural network (CNN) based on transfer learning and conducts a comparative analysis using three different pre-trained Deep Neural Networks (DNNs). The proposed CNN leverages transfer learning techniques with EfficientNetB7, VGG19, and MobileNetV2 to extract features from MRI images. The Adam optimization algorithm is employed to detect and classify different types of BT. The transfer learning approach aids in identifying discriminative features of BT, enabling the classification of various tumor types, such as glioma, meningioma, pituitary, and no tumor. The model's performance is evaluated using metrics like accuracy, precision, F1 score, and recall. Among the tested models, the EfficientNetB7 with transfer learning and ImageNet weights achieved the highest accuracy of 99.47%, outperforming VGG19 and MobileNetV2, which had accuracy of 98.62% and 95.72%, respectively.

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

Performance Comparison of High Accuracy CNNs for Brain Tumor Detection Using Transfer Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Gunjan Jha,
  • Anshul Jha,
  • Eugene B. John

摘要

Brain Tumor (BT) is one of the most life-threatening diseases with a relatively low survival rate. Several diagnostic techniques, including magnetic resonance imaging (MRI), computed tomography (CT), biopsy, cerebral arteriogram, and positron emission tomography (PET), are used to identify BT. With the recent advancements in Deep Learning (DL), radiologists and pathologists can now detect and predict BT with greater accuracy. DL techniques are particularly well-suited for efficiently processing large volumes of image data from various scans. This research introduces a high-accuracy convolutional neural network (CNN) based on transfer learning and conducts a comparative analysis using three different pre-trained Deep Neural Networks (DNNs). The proposed CNN leverages transfer learning techniques with EfficientNetB7, VGG19, and MobileNetV2 to extract features from MRI images. The Adam optimization algorithm is employed to detect and classify different types of BT. The transfer learning approach aids in identifying discriminative features of BT, enabling the classification of various tumor types, such as glioma, meningioma, pituitary, and no tumor. The model's performance is evaluated using metrics like accuracy, precision, F1 score, and recall. Among the tested models, the EfficientNetB7 with transfer learning and ImageNet weights achieved the highest accuracy of 99.47%, outperforming VGG19 and MobileNetV2, which had accuracy of 98.62% and 95.72%, respectively.