A New Method for Medical Image Registration Based on Neural Networks

摘要

During an interactive operation based on 3D image, slices from different model are always reconstructed for getting a high accuracy orientation of the focus. Image registration is a basic step to reconstruction, so the paper presents a new image registration method which constructs a coordinate transform algorithm based on neural networks and acquires ideal medical image through image interpolation transform. The example shows holistic image coordinate transform is realized by improved BP algorithm and image interpolation, and eventually all data can be attained after registration.