A 3-D image registration technique based on fuzzy logic

Abstract

A 3-D multimodal registration approach to bring 3-D medical images acquired by the same or different imaging modalities into spatial correspondence using external markers is proposed and presented. The registration technique estimates the unknown transformation parameters using a fuzzy logic technique. The transformation consists of nine parameters: three magnification parameters (one parameter along each axis), three translation parameters (one parameter along each axis) and three rotation parameters represented by the three Euler angles. Having aligned the two sets of images in the same coordinate system, the developed method is studied quantitatively using Monte Carlo simulation techniques. Consequently, its performance is evaluated by assessing the residuals of the external 3-D points that are used in estimating the registration transformation, the residuals of the registration parameters as well as the residuals of regions in the generated simulated head. The results show that the developed approach is successful in estimating the parameters and the precision is acceptable.