Multi-modal image registration using ordinal features and multi-dimensional mutual information

This paper proposes to use a new feature, namely ordinal feature, for image registration tasks. The ordinal features are extracted by passing through the images through the generalized ordinal filter bank, which effectively encodes the spatial information between neighboring voxels and specific microstructural information in the images. The ordinal feature is integrated with the image intensity to form a two-element attribute vector. Then, four dimensional mutual information is used as the similarity measure. The experimental results show that the proposed method is more robust than the conventional MI-based method and the method using Gabor features. The accuracy is comparable for the three approaches. © 2007 IEEE