Brain segmentation method based on differential geometric information

The invention discloses a brain segmentation method based on differential geometric information. The method comprises the following steps: carrying out standardization processing on a brain tissue image and a brain tumor multi-modal image; based on differential geometric knowledge, introducing a Jacobian determinant and a Laplace operator to emphasize edge information of the image; and after the training quantity is increased by adopting a data enhancement technology, carrying out image segmentation under a neural network model framework. According to the method, training is completed on an IBSR2018 data set and a BraTS2018 data set, and the Dice coefficients are high, which prove that a segmentation framework based on differential geometric information can show good performance in tissue and tumor segmentation. 本发明公开了一种基于微分几何信息的脑分割方法。该方法包括：对脑组织图像和脑肿瘤多模态图像进行标准化处理；基于微分几何知识，引入雅克比行列式和Laplace算子强调图像的边缘信息；采用数据增强技术增加训练数量后，在神经网络模型框架下进行图像分割。本发明在IBSR2018数据集和BraTS2018数据集上完成训练，Dice系数均较高，证明基于微分几何信息的分割框架，能够在组织和肿瘤分割上表现出较好的性