一种卵巢卵泡分割方法、装置、电子设备及存储介质
By combining feature extraction and fusion layers in the ovarian follicle segmentation network, the problem of low accuracy in ovarian follicle segmentation by deep learning algorithms is solved, and high-precision ovarian follicle segmentation is achieved with limited data.
CN116309360BActive Publication Date: 2026-07-17WUHAN ZHONGQI BIOLOGICAL MEDICAL ELECTRONICS
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN ZHONGQI BIOLOGICAL MEDICAL ELECTRONICS
- Filing Date
- 2023-02-16
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing deep learning algorithms cannot effectively utilize limited data in ovarian follicle segmentation, resulting in low segmentation accuracy.
Method used
An ovarian follicle segmentation network, including a feature extractor and a feature fusion layer, is used. The feature extractor extracts high-level and low-level feature maps of the image, and the feature fusion layer performs feature fusion to obtain a fused feature map, which ultimately determines the segmentation result of the ovarian follicle.
Benefits of technology
This method achieves complete preservation of semantic information of ovarian follicles under limited data conditions, thereby improving the accuracy of ovarian follicle segmentation.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116309360B_ABST
Abstract
本发明公开了一种卵巢卵泡分割方法、装置、电子设备及存储介质，方法包括：获取训练完备的卵巢卵泡分割网络，所述卵巢卵泡分割网络包括特征提取器和特征融合层；基于所述特征提取器，对处理后的图像进行特征提取，获得高级特征图和低级特征图；基于所述特征融合层，对所述高级特征图和低级特征图进行特征融合，得到融合特征图；根据所述融合特征图，确定卵巢卵泡图的分割结果。本发明解决了现有技术中使用深度学习算法对卵巢卵泡分割时无法通过有限的数据样本学习全面的特征表达，从而导致卵巢卵泡分割精度低的技术问题。
Need to check novelty before this filing date? Find Prior Art