Image registration is an iterative technique where one image is fitted onto another image of the same structure or location using a combination of rotations and translations in 2D or 3D space. Image registration can be applied to images of the same modality at multiple time-points to quantify local structural change (longitudinal or single-modal registration), or can be applied to two images obtained from different modalities at the same time point to link two signals to the same location (multi-modal registration). At MOIN-CC, image registration procedures are developed for both of these applications.
A registration procedure has been developed to register in vivo micro-CT images in order to produced time-lapse images for the quantification of bone formation and resorption, and to assess the severity of metastatic lesions. The programs have been custom-built in ScancoMedical’s image processing language (IPL) and Matlab.
Using the software Amira, micro-CT images are registered with fluorescent images to link functional information to specific skeletal locations. For example the kinetic uptake of fluorescent bisphosphonates can be linked to regions of high or low bone tissue mineral density. This has been studied in a MOIN-CC project to develop a method for localized bone turnover quantification using binding kinetics.