Scholarlink

<p>Selecting the right embryos for transfer is crucial to improving the success rate of In Vitro Fertilization and Embryo Transfer (IVF-ET). The precise identification and quantification of embryonic cells in the image is the key to the selection of transplanted embryos. However, the variability of the cells’ appearance and their overlap in the images pose a challenge to embryologists. To improve the accuracy and consistency of embryo selection, this paper proposes an improved YOLOv8, YOLOv8-C2fCA, which strategically embeds a Coordinate Attention (CA) into the Cross Stage Partial with Two Fusion (C2f) module of its backbone and neck networks. The resulting C2fCA module significantly improves the model’s ability to extract detailed features from overlapping cell structures. Compared with baseline and attention-based YOLOv8 models, YOLOv8-C2fCA achieved superior results on both private and public datasets, demonstrating higher accuracy, lower computational complexity, and faster inference speed, with only a minimal increase of 0.096&#xa0;M in the number of parameters over the original YOLOv8. This model can be an effective tool for automated detection and counting of embryonic cells, significantly reducing clinicians’ workload.</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

Improved YOLOv8-C2fCA for embryonic cell detection and counting

  • 期刊论文

  • 发表日期：

    2025年4月2日

查看全文
  • Ouafa Talha,
  • Wenju Zhou,
  • Naitong Yuan,
  • Yuan Xu

摘要

Selecting the right embryos for transfer is crucial to improving the success rate of In Vitro Fertilization and Embryo Transfer (IVF-ET). The precise identification and quantification of embryonic cells in the image is the key to the selection of transplanted embryos. However, the variability of the cells’ appearance and their overlap in the images pose a challenge to embryologists. To improve the accuracy and consistency of embryo selection, this paper proposes an improved YOLOv8, YOLOv8-C2fCA, which strategically embeds a Coordinate Attention (CA) into the Cross Stage Partial with Two Fusion (C2f) module of its backbone and neck networks. The resulting C2fCA module significantly improves the model’s ability to extract detailed features from overlapping cell structures. Compared with baseline and attention-based YOLOv8 models, YOLOv8-C2fCA achieved superior results on both private and public datasets, demonstrating higher accuracy, lower computational complexity, and faster inference speed, with only a minimal increase of 0.096 M in the number of parameters over the original YOLOv8. This model can be an effective tool for automated detection and counting of embryonic cells, significantly reducing clinicians’ workload.