Simultaneous registration and intensity normalization of SPECT perfusion images

The analysis of 3D SPECT brain images requires several pre-processing steps such as registration, intensity normalization and brain extraction. Usually, registration is performed before intensity normalization, which requires robust registration methods, such as those based on the maximization of the Mutual Information (MI), which are computationally complex. In this paper we propose using a computationally simple method to perform the simultaneous registration and intensity normalization of SPECT brain perfusion images. The approach, which extends to 3D data a method originally proposed in [1] for 2D photographic images, estimates in alternate steps the intensity normalization parameters and the registration parameters. Our experiments, with real SPECT images, show that the proposed registration method leads to results similar to those obtained by using more expensive algorithms such as those based on the MI criterion.