PET and CT image registration of the rat brain and skull using the AIR algorithm

Abstract

Spatially registered PET and CT images of the same small animal offer at least three potential advantages over PET alone. First, the CT images should allow accurate, nearly noise-free correction of the PET image data for attenuation. Second, the CT images should permit more certain identification of structures evident in the PET images and third, the CT images provide a priori anatomical information that may be of use with resolution-improving image reconstruction algorithms that model the PET imaging process. Thus far, however, image registration algorithms effective in human studies have not been characterized in the small animal setting. Accordingly, the authors evaluated the ability of the AIR algorithm to accurately register PET F-18 fluoride and F-18 FDG images of the rat skull and brain, respectively, to CT images acquired following each PET imaging session. The AIR algorithm was able to register the bone-to-bone images with a maximum error of less than 1.0 mm. The registration error for the brain-to-brain study, however, was greater (2.4 mm) and required additional steps and user intervention to segment the brain from the head in both data sets before registration. These preliminary results suggest that the AIR algorithm can accurately combine PET and CT images in small animals when the data sets are nearly homologous, but may require additional segmentation steps with increased mis-registration errors when registering disparate, low contrast soft tissue structures.