Nonrigid registration for the longitudinal analysis of micro-CT bone datasets

Aims Micro-CT is an established tool to evaluate the bone architecture in small animals. Especially in-vivo micro-CT is very popular, as it allows longitudinal (i.e. over time) follow-up of the evolution in the same animal. This way, changes in the bone structure over time can be assessed. Usually, bone parameters are measured separately in the different images. To be comparable, a similar location is selected in each image using an anatomical descripition (e.g. ‘150 μm proximal to the distal growth plate of the femur’). Several authors have proposed to use automated image registration to improve longitudinal analysis of the micro-CT data. Waarsing etal [1] use rigid registration to detect and visualize changes in the local bone architecture and individual trabeculae over time. More recently, Nishiyama etal [2] have shown that the reproducibility of bone measurements is improved when image registration is used to ensure exactly the same region of interest is chosen in all images.