Scholarlink

<p>The pathologists use histopathology images to identify breast cancer. Under these circumstances, the identification of mitosis in tissues becomes a potent prognostic marker for breast cancer. Mitosis serves as a vital marker for pinpointing areas of tumor aggressiveness and assessing the probability of disease recurrence. The HR-YOLOv8 model is presented in this paper as a highly accurate way to identify regions of breast cancer and detect mitosis. There are two phases to the study. Through the integration of the HRNet blocks into the YOLOv8 backbone, mitosis is identified in the first stage. The MST algorithm locates the breast cancer region in the second stage. The MST algorithm is employed to merge separate nodes, enabling the identification of cancer regions. HR-YOLOv8 is evaluated on MIDOG21, TUPAC16, and MiDeSeC that datasets are specifically focused on breast cancer, using metrics such as accuracy and F1-score for mitosis detection and AUC, sensitivity, and specificity for breast cancer regions. The results obtained from the study show that the proposed model can identify mitosis and recognize breast cancer regions with high precision.</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

HR-YOLOv8: an innovative model to detect mitosis and identify cancer regions in histopathological images

  • 期刊论文

  • 发表日期：

    2025年9月3日

查看全文
  • Nooshin Nemati,
  • Refik Samet

摘要

The pathologists use histopathology images to identify breast cancer. Under these circumstances, the identification of mitosis in tissues becomes a potent prognostic marker for breast cancer. Mitosis serves as a vital marker for pinpointing areas of tumor aggressiveness and assessing the probability of disease recurrence. The HR-YOLOv8 model is presented in this paper as a highly accurate way to identify regions of breast cancer and detect mitosis. There are two phases to the study. Through the integration of the HRNet blocks into the YOLOv8 backbone, mitosis is identified in the first stage. The MST algorithm locates the breast cancer region in the second stage. The MST algorithm is employed to merge separate nodes, enabling the identification of cancer regions. HR-YOLOv8 is evaluated on MIDOG21, TUPAC16, and MiDeSeC that datasets are specifically focused on breast cancer, using metrics such as accuracy and F1-score for mitosis detection and AUC, sensitivity, and specificity for breast cancer regions. The results obtained from the study show that the proposed model can identify mitosis and recognize breast cancer regions with high precision.