Automatic segmentation process of 3D medical images with several neural networks through structured convolution according to the geometry of the 3D medical images

摘要

The present invention relates to an automatic segmentation method for features such as anatomical and pathological structures or instruments visible in a 3D medical image of an object composed of voxels. The method combines N different convolutional neural networks or CNNs with N ≧ 2 to provide an overall software means or configuration having a structured geometry or architecture that is compatible and comparable to the geometry or architecture of the image volume. To provide and to analyze the voxels that form the volume of a 3D image according to N different reconstructed axes or planes, each CNN is assigned to analyze voxels belonging to one axis or plane. It is characterized by being.