A method for the study and evaluation of the performance of medical devices in vivo is disclosed. The medical devices are preferably tampons that are observed and evaluated in the vaginal cavities of women during their menstrual cycles in order to obtain data relative to tampon performance for the improvement of tampon technology. The method preferably uses magnetic resonance imaging (MRI), although other types of imaging techniques may be used. The present invention also resides in a method of MRI analysis in which a three-dimensional surface model image of a medical device such as a tampon or similar feminine hygiene product in vivo is created from two or more two-dimensional cross section images and combined with a second surface model image to produce a single image.