Medical scanning instruments such as computerized tomography (CT) and magnetic resonance imaging (MRI) provide a series of cross sectional images of human body. In general, the number of available 2D image slices is usually insufficient and therefore, the appropriate interpolation methods are usually exploited to estimate missing slices between available image slices. Among previous methods, the shape-based method is very efficient in implementation and achieves reasonable interpolation results, so it has become a widely used method. However, the shaped-based method does not consider the global-geometric changes in shape of the object; thus, it cannot deal effectively with objects with holes, large offsets, or heavy invagination. In this paper, we propose to add several features to control the interpolation of shape-based method. Guided by these extra features, the shape-based method becomes more powerful and it can handle well more general cases for interpolation of medical images that may include complex problems such as the branching, hollow, and invagination.