In computer-aided diagnosis, temporal change over time can be a key piece of information in treatment monitoring and disease tracking applications. Change detection depends on the ability to align the images of the sequence to a common reference, and the ability to build up memory about the image scene over time. In this papef; we will present approaches for model supported image registration and warping developed for change detection in two computer-aided diagnosis applications. The first application is to develop image registration scheme for change detection in mammographic sequence. A key component of this scheme is the site model constructed based on a combination of image analysis procedures. The site model supported multi-step registration leads to a robust change detection derived from the registered mammographic images which will be invaluable in computer-aided diagnosis. The second application is to develop volumetric image warping scheme aimed at lung desease detection and treatment monitoring using 3D images acquired at different breathing stages or different time courses. The model we adopted in this of application is based on the theory of continuum mechanics in order to more accurately account for the non-rigid motion and deformation of the lung itself In addition to the common feature of model-based approach, both applications require the reliable control points in order to obtain a robust registration and warping results. Experimental results on real image data sets show that these two model supported approaches are very promising in quantitatively characterizing the changes in mammographic image sequences and lung CT image volumes.