Signal processing with alpha-stable distributions and applications
Signal processing with alpha-stable distributions and applications
Numerical Methods Using MATLAB
Numerical Methods Using MATLAB
A Level-Set Based Approach to Image Registration
MMBIA '00 Proceedings of the IEEE Workshop on Mathematical Methods in Biomedical Image Analysis
Multi-modal image registration using the generalized survival exponential entropy
MICCAI'06 Proceedings of the 9th international conference on Medical Image Computing and Computer-Assisted Intervention - Volume Part II
Hi-index | 0.00 |
In this paper, based on the SαSdistributions, we design SαSfilters anduse the filters as a new feature extraction method for non-rigidmedical image registration. In brain MR images, the energydistributions of different frequency bands often exhibitheavy-tailed behavior. Such non-Gaussian behavior is essential fornon-rigid image registration but cannot be satisfactorily modeledby the conventional Gabor filters. This leads to unsatisfactorymodeling of voxels located at the salient regions of the images. Tothis end, we propose the SαSfilters formodeling the heavy-tailed behavior of the energy distributions ofbrain MR images, and show that the Gabor filter is a special caseof the SαSfilter. The maximum responseorientation selection criterion is defined for each frequency bandto achieve rotation invariance. In our framework, if the brain MRimages are already segmented, each voxel can be automaticallyassigned a weighting factor based on the Fisher's separationcriterion and it is shown that the registration performance can befurther improved. The proposed method has been compared with thefree-form-deformation based method, Demons algorithm and a methodusing Gabor features by conducting non-rigid image registrationexperiments. It is observed that the proposed method achieves thebest registration accuracy among all the compared methods in boththe simulated and real datasets obtained from the BrainWeb and IBSRrespectively.