Method and device for learning fractional flow reserve based on prior knowledge

The invention provides a method for learning fractional flow reserve based on prior knowledge, and the method comprises the steps: obtaining imaging data and synthetic data of a coronary artery, extracting geometric features of the imaging data, and projecting the geometric features to the synthetic data to obtain fused geometric features; extracting specific information of the blood flow in the synthetic data, and connecting the specific information with the fused geometric features; and predicting the fractional flow reserve value according to the fused geometric features and the specificity information. Real imaging data and synthetic data are fused, the real data are introduced into prediction, the domain gap between synthetic data distribution and real data distribution is made up, and the calculated values of the learned blood flow model and the fractional flow reserve are closer to clinical practice. 本申请提供了一种基于先验知识进行学习血流储备分数的方法，包括：获取冠状动脉的成像数据与合成数据，提取所述成像数据的几何特征，并将所述几何特征投影到所述合成数据得到融合几何特征；提取所述合成数据中的血流的特异