1 INTRODUCTION Medical image processing (MIP) plays very impor- tant role in determination of diagnosis as well as in surgery and therapy. In orthopedic surgery, but also in all other sub-branches of surgery, where the need for creation of customized implants or fixing devices exists, there is a specific requirement for reverse modeling (RM) and reverse engineering (RE) of tis- sue. In general, there are two main directions in re- verse engineering application. The first one is re- lated to reverse engineering of hard tissues like bones, and the second one is related to reverse engi- neering of soft tissues like skin or vascular tissue (Sun et al. 2004a, b). In both application directions, reverse engineering has to deal with tissue geometry. Manufacturing of customized bone’s implant or ex- ternal fixing device in the case of bone fracture re- quires reverse modeling of the bone’s geometry. Manufacturing of soft tissue implant can require cus- tomized scaffold or a woof for tissue growth or res- toring. Thus, it is closely related to the tissue geome- try also. Medical image processing is the starting point of geometry tissue capturing. The radiology image of the tissue represents input set of data for reverse modeling and engineering, which is often called raw data in CAD (computer-aided design) terminology. It is the reason why RE technologies needs to com- municate with MIP closely. The paper reports about a case (geometry of femur) where the results gained from RE can affect MIP and its standards (conven- tions). 2 REFERENTIAL GEOMETRICAL ENTITIES The basic prerequisite for successful reverse modeling of a human bone’s geometry is identification of referential geometrical entities (RGEs). Usually, these RGEs include characteristic points, directions, planes and views. All other elements of the bone’s geometry (curves, surfaces and solids) should be referenced to RGEs. In order to create the robust geometrical model of the bone, which is easy to operate with, one should strive to minimize the set of primary RGEs. Thus, all other geometrical constraints and relations should be based on that minimal set of primary referential geometrical entities. In the same time, this is the approach for successful parametrization of human bone’s geometry. Moreover, the correct identification of RGEs helps RE to be fast and accurate, which is the basic imperative in urgent cases. Referential Geometrical Entities for Reverse Modeling of Geometry of Femur M. Stojkovic, M. Trajanovic, N. Vitkovic, J. Milovanovic Mechanical Engineering Faculty, University of Nis, Serbia S. Arsic, M. Mitkovic School of Medicine, University of Nis, Serbia ABSTRACT: Reverse modeling and engineering of human bone’s geometry is the indispensable part of the process of customized implants and external fixators manufacturing. Because of that, it is very important to remodel geometry of the human bone fastly and accurately. The basic prerequisite for fast and accurate reverse modeling of a human bone’s geometry is identification of the referential geometrical entities (RGEs), that is characteristic points, directions, planes and views. All other elements of the bone’s geometry (curves, surfaces and solids) should be referenced to RGEs. The paper presents idetification of RGEs in the case of specific reverse modeling approach of the geometry of femur, which starts with radiology image input. In ad- dition, the results presented in the paper, suggest some guidelines for the future MIP and CAD software communication. Keywords: Reverse modeling, Reverse engineering, Femur