This paper proposes a new method for multimodality medical image registration using mutual information (MI) as a matching criterion. The method measures the statistical dependence between image intensities of corresponding voxels in both images, assuming this dependence is maximal when the images are geometrically aligned. MI is a general and powerful criterion that does not require assumptions about the nature of the dependence or constraints on image content. The accuracy of the MI criterion is validated for rigid body registration of CT, MR, and PET images by comparison with stereotactic registration solutions, while robustness is evaluated with respect to implementation issues such as interpolation, optimization, and image content. The results demonstrate that subvoxel accuracy can be achieved automatically without prior segmentation or pre-processing, making this method suitable for clinical applications. The paper discusses the theory of mutual information, which measures the statistical dependence between two random variables. It is related to entropy and has properties such as non-negativity, symmetry, and boundedness. The MI registration criterion is applied to multimodality images by estimating joint and marginal distributions from overlapping regions of the images. The criterion is maximized to find the optimal registration parameters. The algorithm involves transforming images using rigid body transformations and estimating image intensity distributions through interpolation methods such as trilinear and partial volume (PV) interpolation. The MI criterion is evaluated by calculating the mutual information between the joint and marginal distributions of the image intensities. The optimal registration parameters are found by maximizing the MI criterion. The paper evaluates the accuracy and robustness of the MI criterion for rigid body registration of CT/MR and PET/MR images. The results show that the MI criterion achieves subvoxel accuracy and is robust to various image degradations such as noise, intensity inhomogeneity, and geometric distortion. The method is also robust to partial overlap between images and can be applied to a wide range of multimodality image registration tasks. The paper concludes that the MI criterion is a powerful and data-independent method for multimodality image registration, suitable for clinical applications.This paper proposes a new method for multimodality medical image registration using mutual information (MI) as a matching criterion. The method measures the statistical dependence between image intensities of corresponding voxels in both images, assuming this dependence is maximal when the images are geometrically aligned. MI is a general and powerful criterion that does not require assumptions about the nature of the dependence or constraints on image content. The accuracy of the MI criterion is validated for rigid body registration of CT, MR, and PET images by comparison with stereotactic registration solutions, while robustness is evaluated with respect to implementation issues such as interpolation, optimization, and image content. The results demonstrate that subvoxel accuracy can be achieved automatically without prior segmentation or pre-processing, making this method suitable for clinical applications. The paper discusses the theory of mutual information, which measures the statistical dependence between two random variables. It is related to entropy and has properties such as non-negativity, symmetry, and boundedness. The MI registration criterion is applied to multimodality images by estimating joint and marginal distributions from overlapping regions of the images. The criterion is maximized to find the optimal registration parameters. The algorithm involves transforming images using rigid body transformations and estimating image intensity distributions through interpolation methods such as trilinear and partial volume (PV) interpolation. The MI criterion is evaluated by calculating the mutual information between the joint and marginal distributions of the image intensities. The optimal registration parameters are found by maximizing the MI criterion. The paper evaluates the accuracy and robustness of the MI criterion for rigid body registration of CT/MR and PET/MR images. The results show that the MI criterion achieves subvoxel accuracy and is robust to various image degradations such as noise, intensity inhomogeneity, and geometric distortion. The method is also robust to partial overlap between images and can be applied to a wide range of multimodality image registration tasks. The paper concludes that the MI criterion is a powerful and data-independent method for multimodality image registration, suitable for clinical applications.