Title :
Projection-based registration of radiological images1
Author :
Eroglu, Cigdem ; Cetin, A. Enis
Author_Institution :
Dept. of Electr. & Electron. Eng., Bilkent Univ., Ankara, Turkey
Abstract :
This paper introduces a projection-based method for registration of radiological images. The proposed method estimates the scale, translation and rotation parameters which define the global affine motion of an image with respect to a reference image. Using these estimated parameters, radiological images can be properly re-aligned to help diagnosis. The proposed method tries to match the horizontal and vertical projections of the image to be registered with the corresponding projections of the reference image in a hierarchical manner. After coarse estimates of the parameters are obtained using low resolution images, higher resolution images are utilized for the refinement of the parameter estimates. The experiments carried out on Computed Tomography (CT) and Magnetic Resonance (MR) images show that the performance of the algorithm in estimating the affine motion parameters is encouraging.
Keywords :
affine transforms; biomedical MRI; computerised tomography; image motion analysis; image registration; image resolution; medical image processing; parameter estimation; radiology; affine motion parameters; computed tomography; image resolution; magnetic resonance images; parameter estimates; projection-based registration; radiological images; reference image; rotation parameters; scale parameters; translation parameters; Accuracy; Clocks; Computed tomography; Estimation; Image resolution; Radiation detectors; Signal processing algorithms;
Conference_Titel :
Signal Processing Conference (EUSIPCO 1998), 9th European
Conference_Location :
Rhodes
Print_ISBN :
978-960-7620-06-4