关键词: 互信息, 医学图像配准, 部分体积插值法, 加权熵

Abstract: The medical image registration by mutual information has been accepted as one of the most accurate and highly-
automated methods.For the mutual information calculated by partial volume interpolation method and Shannon entropy，certain local extremums are inevitable，which may lead to inaccurate registration.Weighted entropy instead of Shannon entropy to compute the mutual information is proposed in this paper，and it has been used in medical image registration experiment.The experimental results show that the method can smooth the local extremums of mutual information and reduce the possibility of wrong medical image registration.

Key words: mutual information, medical image registration, partial volume interpolation, weighted entropy