Scholarlink

<p>Ovarian cancer is a fatal health condition and one of the most common women’s diseases that affects millions of women around the world. Ovarian cancer is a prominent source of gynecological cancers, and its early diagnosis is required to improve patient survival. In this study, the performance of 5 pre-trained Convolutional Neural Networks (CNNs), i.e., VGG16, VGG19, ResNet50, MobileNet, and DenseNet121, is compared for the classification of ovarian cancer from histopathological images. Additionally, traditional machine learning classifiers like K-Nearest Neighbors (KNN), Support Vector Machines (SVM), and Gaussian Naive Bayes were compared with the models. The histopathological images were processed using Wiener filtering, Contrast Limited Adaptive Histogram Equalization (CLAHE), and data augmentation techniques. All the models were tested with metrics such as accuracy, precision, recall, and F1-score. DenseNet121 yielded the highest accuracy of 95.3%, indicating better performance in cancer pattern identification compared to other CNNs and traditional classifiers. The findings suggest the prospective application of deep learning models, in this case, DenseNet121, for improving diagnostic practices in ovarian cancer. The work enhances histopathological image analysis and provides important insights for incorporating machine learning strategies into clinical environments for the support of early detection and effective treatment planning. Unlike existing studies, this work combines multiple real-world datasets. It delivers a robust comparative evaluation across traditional and deep learning techniques, offering valuable insights for practical deployment in clinical settings.</p>

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

Deep learning-based ovarian cancer detection from histopathology images

  • 期刊论文

  • 发表日期：

    2025年8月2日

查看全文
  • Thulasi Bikku,
  • Jeevana Jyothi Pujari,
  • K. P. N. V. Satyasree,
  • Srinivasarao Thota,
  • S. Joseph

摘要

Ovarian cancer is a fatal health condition and one of the most common women’s diseases that affects millions of women around the world. Ovarian cancer is a prominent source of gynecological cancers, and its early diagnosis is required to improve patient survival. In this study, the performance of 5 pre-trained Convolutional Neural Networks (CNNs), i.e., VGG16, VGG19, ResNet50, MobileNet, and DenseNet121, is compared for the classification of ovarian cancer from histopathological images. Additionally, traditional machine learning classifiers like K-Nearest Neighbors (KNN), Support Vector Machines (SVM), and Gaussian Naive Bayes were compared with the models. The histopathological images were processed using Wiener filtering, Contrast Limited Adaptive Histogram Equalization (CLAHE), and data augmentation techniques. All the models were tested with metrics such as accuracy, precision, recall, and F1-score. DenseNet121 yielded the highest accuracy of 95.3%, indicating better performance in cancer pattern identification compared to other CNNs and traditional classifiers. The findings suggest the prospective application of deep learning models, in this case, DenseNet121, for improving diagnostic practices in ovarian cancer. The work enhances histopathological image analysis and provides important insights for incorporating machine learning strategies into clinical environments for the support of early detection and effective treatment planning. Unlike existing studies, this work combines multiple real-world datasets. It delivers a robust comparative evaluation across traditional and deep learning techniques, offering valuable insights for practical deployment in clinical settings.