Research on Medical Image Registration Based on Graphic Neural Network Reinforcement Learning
Zhejia Dong · Journal of Physics Conference Series · 2020
Abstract On the basis of traditional medical image registration methods, aiming at the shortcomings of rigid body images, sequential rigid body images and non-rigid body images registration methods, this paper proposes a new method of medical image registration application based on graphic neural network reinforcement learning. Reinforcement learning adjusts the strategy from the actual system learning experience, and it is a process of gradually approaching the optimal strategy, mapping the image from high-dimensional gray space to low-dimensional feature category space; Then, the elastic registration of mutual information images is carried out in the feature class space. The advantage of this method is that it does not need extra supervision information, which effectively saves manpower and material resources. So as to make an accurate diagnosis or make a proper treatment plan.