摘要:回顾了神经网络理论发展的历史和现状。在此基础上，介绍并讨论了20世纪如年代神经网络研究的一些新进展。根据神经网络研究的特点，对人工神经网络今后的发展前景作了一定的评述，

人工神经网络的研究与发展必将对现代科学技术产生深远的影响.配准中基于像素和基于特征两类主要方法的优缺点 深入研究了互信息理论在医学图像配准领域的应用和实现机理以及互信息作为相似性测度的优缺点。本文着

重深入研究了Renyi广义互信息及其在图像配准中的作用，并针对互信息计算的复杂性，引入了广义近邻图嫡估计理论，并成功应用于多模医学图像配准中。

-Abstract: The review of the history and current status of the development of neural network theory. On this basis, some new developments in the 20th century, such as the age of neural network research is presented and discussed. According to the characteristics of neural network research, made some comment on the future prospects for the development of artificial neural network, artificial neural network research and development is bound to have a profound impact on modern science and technology. Registration based on the pixel-based features of the two main types The advantages and disadvantages of the method -depth study of the mutual information theory in the field of medical image registration and achieve mechanism and the advantages and disadvantages of mutual information as a similarity measure. The article focuses on in-depth study the Renyi generalized mutual information and its role in image registration, and the introduction of mutual information calculation complexity, the t