病理图像虚拟染色方法、电子设备及存储介质
By combining adaptive preprocessing, feature compression networks, and Brownian bridge diffusion models with adaptive constraints, the batch effect and cell adhesion problems in virtual staining technology were solved, achieving efficient and stable virtual staining of pathological images and meeting the real-time needs of clinical practice.
CN121962312BActive Publication Date: 2026-07-17XIANGYA HOSPITAL CENT SOUTH UNIV
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIANGYA HOSPITAL CENT SOUTH UNIV
- Filing Date
- 2026-04-03
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing virtual staining technologies suffer from batch-to-batch staining instability, cell adhesion issues, and slow inference speed when processing clinical HE slides to IHC, failing to meet the real-time requirements of clinical practice.
Method used
Efficient and robust virtual staining of pathological images is achieved through adaptive preprocessing, feature compression network mapping, Brownian bridge diffusion model transformation, and adaptive constraint guidance.
Benefits of technology
It significantly improves the stability and quality of virtual staining results, reduces computational complexity, meets the needs of real-time clinical interaction, and the generated virtual IHC images are consistent with pathological authenticity.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN121962312B_ABST
Abstract
本发明公开了一种病理图像虚拟染色方法、电子设备及存储介质，涉及医学图像处理技术领域。所述方法包括对HE染色病理图像自适应预处理得标准化图像块；将其输入特征压缩网络以映射至低维潜空间，得第一潜空间特征向量；以第一潜空间特征向量为初始状态，输入布朗桥扩散模型，在低维潜空间反向扩散生成第二潜空间特征向量；在反向扩散过程中，基于标准化图像块计算组织密度感知参数并动态调整多尺度生物学约束损失权重，用调整后损失计算梯度引导迭代方向；将第二潜空间特征向量解码输出虚拟IHC图像。本发明结合潜空间压缩与布朗桥扩散实现亚秒级推理，通过自适应生物学约束解决了高密度细胞粘连，生成高保真度的虚拟IHC图像。
Need to check novelty before this filing date? Find Prior Art