Efficient Multi-Modal Least-Squares Alignment of Medical Images Using Quasi-Orientation Maps

In image registration, similarity metrics are used to determine the optimal alignment between two images. A common metric used for judging image similarity is the weighted sum of squared differences (SSD) cost function. Recently, it was demonstrated that the evaluation of the SSD cost function can be performed efficiently using the Fast Fourier Transform (FFT) to determine the optimal translation between two images based on pixel intensities. This paper extends this efficient approach by introducing the concept of quasiorientation maps as features into the alignment framework. This feature-based method is invariant to intensity mappings, making it suitable for aligning medical images acquired with different modalities. Experimental results demonstrate overall multi-modal image alignment performance to be superior to that of