CT and MRI Brain Images Registration for Clinical Applications

Abstract

Registration is particularly challenging when fusing imagery from different sensors with different resolutions. Ridges are useful geometrical features for image registration. In this paper, first the performance of Gaussian and Butterworth high pass filters is analyzed on sharpening of images, then ridges are detected from CT and MRI brain images by using scale space primal sketch, and finally cross correlate the two 3D CT and MRI volumes for registration. Results are presented on number of images. The visual comparison shows that filters at cutoff frequency 100 MHz shows better quality. The comparative analysis show that Butterworth filter of order four yields more sharper images as compared to Gaussian filter. Finally, visual results after registration of MRI and CT images of the brain are presented, which can be useful in planning certain types of neurosurgical and ENT surgical procedures