PDF995, Job 2

Technology and Health Care 8(2000)1-11 THC281 IOS Press

1

Design and manufacturing of cranioplasty implants by 3-axis cnc milling L.C Hieu a, E. Bohez a, J.Vander Sloten a,b, P. Oris a,b, H.N.Phiena, E.Vatcharaporn aand P.H.Binhc a

School of Advanced Technologies, Asian Institute of Technology, Thailand Division of Biomechanics and Engineering Design, Katholieke Universiteit.Leuven, Belgium c Department of Neurosurgery, General Central Military Hospital 108, Hanoi, Vietnam b

Received 20 February 2002 Abstract. Although various techniques and materials have been used for making cranioplasty implants, personalized cranioplasty implants are high in cost because of expensive materials and production technology, long design and manufacturing time, and intensive labor use. This research was a part of our research project in ASEAN countries to investigate feasible technical solutions of minimizing the implant cost based on available production technologies in the region. The use of 3-axis CNC (Computer Numerical Control) milling techniques for making molds to fabricate PMMA implants was successfully investigated. With the development of a design support program bridging between Computer Aided Design (CAD) and Medical Image Processing (MIP) system, the time for geometrical modeling of implants and molds was reduced to half a day. The machining time to complete a mold was about 5 to 6 hours; and it took maximal 2 hours to fabricate an implant with selfcuring PMMA and 3 and half hours for fabricating an implant with heat-curing PMMA. The cost of implants is acceptable for the ASEAN region.

Keywords: Cranioplasty-implant-CAD/CAM

1. Introduction With the advent of computed tomography (CT) in the early 1970s, precisely cut sections of the human body were obtained; and nowadays CT & MRI (Magnetic Resonance Imaging) scanning has been known as one of the best methods of getting the information for medical diagnosis and planning. Based on CT or MRI scanning data, three-dimensional (3D) models of an anatomical structure of the human body are constructed, and there are many applications derived from these 3D models. Cranipplasty is the method of treatment of skull defects; it is required to protect underlying brain, correct major aesthetic deformities, or both [22]. In the field of cranioplasty, the use of the state-of-the art design and manufacturing technologies (Computer Aided Design, Computer Aided Manufacturing CAD/CAM, Rapid Prototyping, etc.) and 3D modeling of anatomical structures, have been used mainly used for two purposes: (i) making 3D physical models (biomodels)[6,11,20,21] and (ii) implant design [7-10]. Biomodels are used for preoperative planning, used as a template for preparing implants [20] and used as a master implant to make a mold from which the implant is fabricated [3]. A variety of alloplastic materials have been used for making cranioplasty implants [1,2]. Titanium and Polymethyl methacrylate - PMMA (acrylic) are the most commonly used. Although various techniques [3,7-14] have been used for making implants, personalized implants are high in cost because 0928-7329/02/$8.00 2000 –IOS Press. All right reserved

2

L.C.Hieu et al./ Design and manufacturing of cranioplasty implants by 3-axis cnc milling

of expensive materials as well as design and manufacturing technology, long design and manufacturing time and intensive labor use [3,6,7]. That inspired us to investigate technical solutions to make PMMA cranioplasty implants by 3-axis CNC milling for the ASEAN region where the use of 5-axis CNC milling, rapid prototyping (RP) and hydroforming are expensive and not always available. The complex skull geometry leads to the difficulties in using 3-axis CNC milling techniques to make the molds for fabricating implants; especially the problems of over and under cuts normally occur in the machining process. By introducing the ruled free-form splitting surface and multiple parts for the molds, the over and under cuts could be avoided. Moreover, the contact between the bone and implants should be taken into account during the design process because it influences very much in the problem of the over and under cuts. Related to geometrical modeling of the cranioplasty implant, geometry of implants is determined through three surfaces: outer, inner and contact surface; a contact surface is the one that contacts with bordering surface of the bone window (defect window); and inner (outer) surface of an implant are respectively corresponding to the inner (outer) wall of the cranial bone. The modeling process of craniolasty implants is started with the use of CT scanning data as the input, which is then segmented to separate the bonny structure from the soft tissue in MIP software through thresholding techniques. The bonny geometry in the form of polylines (IGES - Initial Graphics Exchange Specification) or solids (STL - STereoLithography) files is then interfaced to CAD/CAM software for the detailed geometrical modeling of an implant. In order to convert the raw bonny data resulted from MIP into the one, which is compatible with CAD/CAM principles, intensive labor use and design skills are required to separate the inner and outer contours of the skull, data reduction and reformatting the design data. We developed the program bridging between MIP and CAD for automatically solving the mentioned design data processing chain. In this paper, the method for geometrical modeling of the cranioplasty implants based on hybrid surface/solid modeling, and the solution of designing the mold for fabricating PMMA implants by 3-axis milling, were presented. The problems related to the design data processing chain are emphasized. 2. Material and method The patient CT scanning data was processed in the commercial MIP software – MIMICS (Materialise NV, Belgium). The bonny structural geometry of a skull outputted from MIMICS were in the form of 3D solid models: STL files, or in the form of contours: SSL (StrataSys Layer interface) and IGES files. These STL, SSL and IGES files were then used as the basic data for further design steps. For the detailed modeling of implants/molds and tool path generation, UG (Unigraphics Solutions) and MasterCAM (CNC software inc.) were used. 2.1. The program bridging between MIP and CAD The program was developed and integrated into the design process to make a bridge between MIP and CAD. This in-house developed program was used for the following purposes: (i) filtering noises from IGES and SSL-files, and extraction of outer wall contours of the neurocranium; and (ii) optimizing the data structure of outer wall contours (directions, number of control points and their distribution). The IGES and SSL files of the cranial bonny contours, which are directly outputted from MIMICS, are used as the inputs for further design data processing steps. However, when modeling an implant, 3D skull models are used for surgical planning; and it is, therefore, suggested that a 3D solid skull model in the form of


3

Fig 1. The twisted and waved surface problem (A, B, C), and the surface of “evenly” distributed control points in U-V directions (D).

STL file should be used; this STL-file is then sliced for the SSL file in RP slicing software such as Magics RP (Materialise NV, Belgium), and finally this SSL file is used as the inputs for the program bridging between MIP and CAD. In addition, the reason of using STL files is that mirror-imaging techniques are fundamentally used for the implant modeling; and this is easily done on STL files. Moreover, by slicing the STL file, the bonny contours of the skull can be reconstructed in a proper direction for geometrical modeling of the outer wall surface of a skull. In fact, the bonny contours extracted directly from the CT scanning data are not always satisfied for reconstructing the outer wall surface of a skull, because the scanning direction is not always compatible with the required lofting direction when constructing the surface from these bonny contours. The resulted IGES and SSL files contain large amount of data and a lot of noises due to complex geometry of the skulls and control points used to present the bonny contours; these files need to be filtered. Since only the outside surface of the neurocranium is needed for geometrical modeling of implants (molds), the bonny outer wall contours needs to be separated from the inner ones; the cranial outer surface was constructed through these bonny outer contours or U-V mesh of control points defined by contours. The twisted and waved surface problem practically always occur when modeling outer wall surface of the neurocranium (skull) in most of the CAD/CAM systems as shown in Figs 1(A), (B) and (C). The unevenly distributed control points along the curve and a different starting and ending point arrangement may lead to a waved surface (Figs 1(A) and (B)); and the contours with different directions lead to a twisted surface (Fig. 1(C)). Beside the twisted and waved surface problem, a large file size and a high degree of mathematical description of the surfaces due to a large number of control points make very difficult for geometrical modeling of an implant/mold, generating an optimal tool paths (CNC milling), and slicing STL files (RP). Therefore, for all contours, directions, number of control points and their distribution need to be reformatted to be optimized ones. These optimized data are then imported into the end-use CAD software for detailed design of an implant (mold). Fig. 1(D) shows the surface of an optimal data structure, it was interpolated from the contours with the same number of control points, which are “evenly” distributed along the curve, and good arrangement of starting and ending point. The

4


selection of end-use CAD/CAM software for the further design and manufacturing steps is dependent on complexity of the implant (mold) models, data transfer among packages, tool path generation methods and modeling capabilities of the packages. 2.2. An overall approach for the design of an implant NURB (Non-Uniform Rational B-spline) based surface and solid as well as surface/solid hybrid modeling were used for geometrical modeling of implants and molds for fabricating implants. The overall approach for geometrical modeling of an implant is composed of two steps: (i) reconstructing the intact skull (a part of the skull), which covers the defects; (ii) based on the intact skull (a part of the skull) and bone window of the defect, the contact as well as inner and outer surface of an implant is constructed. Techniques used to reconstruct the intact skull (part of the skull) that covers the defect window were fundamentally based on mirror imaging techniques in which non-defected part from contra-lateral side of the skull was used [14,15]. For patients with the skull of a big asymmetry or cases of defects beyond the midline of the skull (called as beyond-midline defects), mirror-imaging techniques cannot be applied; the other skulls are used as references for the design, or the designer has to manually draws CAD entities (lines or points) from which an implants (mold) is modeled. These methods take a lot of time and skills or may result an unexpected surface because of not well-structured design data (Figs 1 (A), (B) and (C)). In order to minimize the design time as well as to solve problems of beyond-midline defects and big asymmetry, a design-support database was constructed in the form of a skull template (concretely, the template of Vietnamese skulls) by analyzing a set of skull’s CT data (more than 200 skulls). This 3D model of the skull template was determined by slice contours, which were calculated by “averaging” outer slice contours of skulls. The design database was used in the form of common CAD entities (point cloud, curves and surfaces) with different interfaces to CAD software. Scaling in X, Y and Z direction or offset options are applied on a skull template to generate the best-fit reference skull for the design. The bordering surface of a (defect) bone window, which will contact with a contact surface of an implant, has typical shapes as presented in Figs 2(A), (B) and (C). A contact surface of an implant is formed by a cutting surface CS (Fig. 2(D)). The following procedure was used for implant modeling (Figs 2(D) and (E)). Firstly the outer wall surface of the neurocranium is constructed from its outer wall contours of the intact one. The outer wall surface is then thickened to a distance of the implant thickness to make a solid (S) that contains an implant. The cutting surface CS is formed by contact contours, which are interactively generated based on the bordering surface of a defect (see the section of “Contact surface design”). The implant model is finally generated through a Boolean subtraction operation between a solid S and a cutting surface CS; Fig. 2(E) shows the result of a Boolean subtraction operation and the implant model was moved out of the solid S. If hybrid surface/solid modeling is not available in the working CAD software, a cutting surface can be constructed as a solid entity by thickening or extruding a surface to a small distance. 2.3. Noise filtering and extraction of outer wall contours The IGES or SSL files was filtered to reduce the amount of data as well as noises. There were two methods used for data reduction. The first method was to base on the out of line distance. If an evaluating point lies within the specified distance of the chord between the previous and the next reading point, it is omitted. The second refinement method was to base on a data reduction criterion reported by M. Fujimoto and K. Kariya [16]; about 80% of the data can be reduced without compromising the accuracy.


5

Fig.2. Typical bordering surfaces of a bone window of the cranial defects (A, B, C); and generation of an implant model based on solid, hybrid surface/solid and NURB modeling techniques (D, E).

The method presented by J. Vander Sloten [17] for separating inner and outer wall of the bone could be applied. However, for the case of cranioplasty, the problem becomes easier because only cranial area is taken into account for the design. The inner and outer wall contours are separated by relying on intersection points among the contours and straight lines (cutting lines) crossing the contours, the contour goes through the farthest intersection point calculated from the local points, which are on cutting lines and inside a contour, is the outer one. 2.4. Optimizing the data structure of contours for outer wall surface construction The line scanning method was used to format outer wall contours. All the contours are reformatted to contain the same number of control points and direction. For all contours, control points are “evenly” distributed in U-V direction of the U-V control point mesh, which establishes the outer wall surface. The starting (or ending) points of all the contours are located on the same plane (Fig. 1(D)) that is perpendicular to the contour (slice) plane and goes through the reference center (RCP) point. The RCP is considered as the point located at “center” of the contours in both X and Y direction; it is relatively a center of the width and length of the head at a certain slice. Assumed that a local reference system located at RCP is OL. A scan line L is the one that goes through OL. By rotating a scan line L around OL with a stepping angle β, intersection points between L and a slice contour will be collected and these collected points form a new contour; this procedure is repeated for all the slices. By this way, all contours contain the same number of control points and direction; and these control points are “evenly” distributed in U-V direction. During the implant design process, partially cranial area are normally used instead of using a full neurocranium; by controlling the starting and ending angle as well as number of slice contours, the interested cranial part is constructed. The error analysis about the number of control points used to construct the slice contours was implemented for a typical skull as shown in Fig. 3. The analysis was done for 19 layers from bottom up, with the starting point g (glabella point) and layer thickness of 5 mm. The total number of control points at each outer wall contour of the slices was arranged from 405 to 539. For each outer wall contour, 36, 72


Maximum Errors in [mm]

6

1 240 Points 72 Points 36 Points

0.8 0.6 0.4 0.2 0

Average Errors in [mm]

Slice 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

240 Points 72 Points 36 Points

1

3

5

7

9

11

13

15

17

Slice

Fig. 3: Maximum and average error analysis about number of control points for slices of a typical skull

and 240 control points were collected according to stepping angles β, which is equal to 1.5, 5 and 10 degrees respectively. The errors were calculated through the distance from original points to the cubic curves (splines) that were interpolated from collected control points. When the contour was interpolated from 36 control points, although the maximal error was up to 0.8 mm, the interpolation curve was in good fit to most of the points and the average error was less than 0.14 mm, which is acceptable for the cranioplasty implant design. Of course, the designer can flexibly select the number of control points for the best fitness; it depends on the position of defects as well as clinical and aesthetic constraints. 2.5. Contact surface design Because the implants are fixed to the skull by mini plates or wires; therefore practically we can control 6 DOF (Degrees Of Freedom) of the implant by using at least 3 contact areas distributed around the bone window. In fact, a completely fit contact between the implant and bone window is not necessarily required. The designer, therefore, can flexibly select the contact positions on the bordering surface of the bone window, and then he or she mainly concentrates on designing the contact areas at these critical positions; the remaining areas may get less attention than the selected ones; by this way, the time for designing an implant’s contact surface is remarkably reduced. Moreover, when CNC milling techniques are used for machining the mold used to fabricate implants, the problem of over (under) cuts can be avoided if contact positions are properly selected. Two types of a contact surface, line and surface, (Figure 4) were used: (i) Line contact: the contact between bone and implant is the line as shown in Figs 4(A) and (B) by point P1 in the cross-section view. The curve P2-P1-P2’ of a surface contact in the cross-view is a straight line. (ii) Surface contact: the contact between bone and implant is the surface as shown in Fig. 4 (C) by the curve P2-P1-P2’ in the cross-section view. The contact surface is generated by lofting three


7

Fig. 4. Line (A, B) and surface (C) contact between bone and an implant

contact contours that go through positions P2, P1 and P2’ on a bordering surface of the bone window. In some cases, the ruled surfaces (P2-P3 and P2’-P3’) are used to connect with the contact surface (P2P1-P2’) to make a whole bordering surface of the implant. The contact contours {Pi, i=1,2,2’} are interactively generated based on the bordering surface of a defect. Contours {P2, P2’} in a line contact and contours {P3, P3’} in both contact types are constructed by scaling and/or offsetting a curve P1. Because, in the surface contact, the load transferring from an implant to the skull is distributed on the surface area, the concentrated stress on bonny border is reduced; and the surface contact is, therefore, better than the line contact in term of loading transfer. However, the machinability by 3-axis CNC milling for the surface contact is less than the one for the line contact because of over (under) cut problems. In the case of using 3-axis CNC milling to make a mold for fabricating the cranioplasty implant, depending on the complexity of the implant (mold)’s geometry, the surface or line contact is selected so that the over (under) cuts are avoided. 2.6. Design and manufacturing of the mold The geometrical model of a mold for fabricating PMMA implants is fundamentally generated based on an implant model. The method of generating the lower and upper cavity is shown in Figs 5(A) and (B). Assumed that ξ1 is the edge contour, which is determined by outer and contact surface of an implant. A free-form splitting surface of the mold is a ruled surface R constructed by contours ξ1 and ξ2 in which ξ2 is an offset or scaled contour of ξ1. An upper (lower) cavity of the mold is constructed by using Boolean subtraction operations among a solid implant model, ruled surface R, and a solid cylinder (box). With the use of a ruled free-form splitting surface and balancing the positions P2 and P2’ as well as contact point P1 in the line contact (Fig. 4), the over and under cuts when machining the mold are remarkably minimized. As mentioned, the complex geometry of an implant leads to difficulties to machine the mold for fabricating PMMA implants by 3-axis CNC milling because of the over (under) cuts remained; and expensive 4 (or 5)-axis milling machines normally need to be used. When 3-axis CNC milling is used, with the use of a free-from splitting surface R, the over (under) cuts can only occur in the lower cavity (Fig.5(C)). Two types of the mold were investigated: type I and type II as shown respectively in Figs 5(A) and (B). For the type II, small ball-end mill tools were used at the finishing operation to reduce the under cuts. According to our experiments, ball end mill tool with the diameter of 2 mm could avoid most of the under cuts; and this was also simulated in the UG software. When a type II could not be used, a type I was used; the only disadvantage of the type I is that it is more material usage than the type II. However,

8


Fig. 5. (A): A mold for fabricating PMMA implant-type I. (B): A mold for fabricating PMMA implant- type II. (C): Over (under) cut problem. (D): Patient with a defect and designed implant. (E): The implant master (biomodel) (a) made by rapid prototyping technique (Z-Corp. Technology, USA); the biomodel (b) and PMMA implant (c) fabricated by the mold that was machined by 3axis CNC milling (DM GmbH, Germany). (F): The lower and upper cavity of the mold, and PMMA implant on the lower cavity of the mold.

bigger tools could be used for the type I at draft and finishing operations (ball and end mill tools with the diameter of 6 to 10 mm), the total machining time to complete the mold for the type I and type II was almost the same. The materials used for making the mold were hard wood resins and plastics. 3. Result The implants made by different techniques were shown in figs 5 (D), (E) and (F). With the use of the in-house developed program for bridging between MIP and CAD, and based on (NURB) free-form surface, solid and hybrid surface/solid modeling, the time for completing the design of an implant and a mold was reduced to half a day. The 3D model of a designed implant was checked in MIP (MIMICS) is shown in figure 5 {D}. The machining time by 3-axis CNC milling to complete the mold (both upper and lower cavity) was about 5 to 6 hours; and it took maximal 2 hours to fabricate an implant with self-curing PMMA and 3 and half a hour for fabricating an implant with heating PMMA. 4. Discussion and conclusion Today, in the field of cranioplasty, titanium and polymethyl methacrylate - PMMA (acrylic) are the most commonly used for making implants. The advantage of acrylic includes radiolucency that is necessary for patients requiring CT or MRI follow-up, poor thermal and electrical conductivity, low cost, easy casting, demonstrated long-term biocompatibility [1,3]. The drawbacks of acrylic are the presence


9

of free monomer, raising the possibility of carcinogenic effects, and the development of allergic reactions, but such monomer toxicity is short-lived; the failure rate of PMMA is ranged from 1.7 to 12 % [1,4,5]. Prefabrication of PMMA implants in molds avoids heat development and monomer-emission in the site of the defect [8]. Titanium has the advantage of being biologically inert; the infection rate is small, it is of under 2% [5]. The disadvantages of titanium include high cost, difficulties in molding, casting or milling, the generation of artifacts in subsequent CT and MRI, high thermal and electrical conductivity [1,3]. Construction of the first craniofacial foam models (biomodels) by 4-axis CNC milling techniques were reported by J.T. Lambrecht and F. Brix [18]. An indirect method of using CAD/CAM to make cranioplasty implants were investigated by some authors in which biomodels of an implant or skull (or a part of the skull) were made; these biomodels were then used as templates for shaping implants [6,19] or used as implant masters to make molds for fabricating implants [14]. This approach leads to a lack of precision and wastes the advantages of CAD/CAM [7]. The direct CNC milling of the mold for fabricating PMMA implants was first introduced in 1995 by H. Eufinger [7] for one case study in which the surface modeling technique was used to model the implant; methods of designing the mold for 3-axis CNC milling was not presented. Although powerful CAD functions exist in most state-of-the art CAD/CAM software today, the complex geometry of the skulls requires intensive labor use and skills to solve the problems of large data, noises and separation of the inner and outer contour. In addition, computer manipulation of CT data were at times painstaking, and 1 to 2 hours of data manipulation was often required if the patient’s anatomy was not symmetrical [3]. With development of the program bridging between MIP and CAD, the raw data outputted from MIP is automatically reformatted to be well structured and practically compatible with geometrical modeling methods in CAD/CAM systems, large amount of data is reduced, and finally the design time and intensive labor use was minimized. For the patients with the defects of being beyond the midline of the skull, mirror-imaging techniques can not be applied for the modeling an implant, the construction of a design support database showed a very useful approach; it helped to reduce lots of time in finding the skull references to match with the patient skull. Thank to the progress of MIP and CAD/CAM industry, today, surface, solid as well as hybrid solid/surface modeling techniques coexist in an unified design environment, the geometrical model of an implant or molds and optimal milling tool paths are quickly generated with the highest accuracy and reliability. The design model of an implant is transferred back to MIP software with a complete 3D view for preoperative planning and design checking; a better communication between surgeons and designer is established. According to our investigation, the use of 3-axis CNC milling for making PMMA implants has several advantages over the method of using RP technology as the follows: (i) More accuracy; (ii) Directly making the mold for fabricating the implants; (iii) Shorter manufacturing lead-time; and (iv) Cheaper and available in the ASEAN region. The method presented can potentially be applied for making the mold to fabricate implants made of bone cement or hydroxyapatite [12]. The production time of RP medical models (Stereolithography) is about 3 times as long as that of CNC milled models [23]. The drift-away and shrinkage problems as well as layer thickness limitations lead to less accuracy in the RP models compared to the milled ones. In the method of using RP technology to make PMMA implants, the RP biomodel of an implant is used as a positive to make the silicon or dental stone mold from which the PMMA implant is fabricated; these steps add more inaccuracies and production time. Technical precision of up to 0.05 mm can be obtained by CNC milling techniques, and it is up to 1/200 mm possible in modern CAD/CAM systems [8]; therefore the implant fabricated by CNC milled mold is clinically reliable enough for cranioplasty applications; in fact, the error of smaller

10


than 0.5 mm is sufficiently accurate; for implants at the occipital area and the nearby ones, the error up to 1 mm can be acceptable. For the countries with limited resources and low income, the cost is the most important factor for the selection of design and manufacturing technologies; sometime it overrides the standard selection criteria in both technical and clinical point of views. The RP medical models (Stereolithography) are 2-3 times more expensive than milled models [23]. D' Urso P. S [3] reported that if RP technology was used, the implant cost was of approximately $1300 US per case ($1000 for the SL biomodel and $300 for the acrylic casting). According to M. Perry [6], to make a biomodel by RP technology, it took 8 to 12 hours (12 to 24 hours on the SLA machine [21]) and it was expensive (at least $950). H. Eufinger [7], reported that the whole process of making titanium implants by machining titanium blocks took at least 3 days, and the high expense for construction and fabrication (approximately $4000 per implant, covering staff and machines); the cost of the systems reached about $100,000 for the CAD system (hardware and software) and about the same amount for an adequate milling machine. The mentioned prices are too expensive for ASEAN countries. In addition, RP machines are not always available in our region, and quite expensive; it is at least $225 for one implant biomodel with the size of about 150x160 mm and 5 mm thick, and it is commercially about $90 for one machine working-hour, not included operation cost. In our investigation, with the use of 3-axis CNC milling, it took totally about 3 days to complete the implant; and the total cost of approximately less than $300 for one PMMA (Orthocryl, Germany) implant is acceptable for ASEAN countries. Acknowledgements We thank Thammasat Hospital, Thailand, for case studies and the following hospitals in Vietnam for providing us the CT patient data: Tien Giang hospital, Cho Ray hospital, Vietsovpetro Medical Center, and Military Hospital 175; and Flemish Interuniversity Council (VL.I.R), Belgium, is acknowledged for the financial support. References J. Beumer, D.N. Firtell and T.A.Curtis, Current concepts in cranioplasty, The journal of prosthetic dentistry 42 (1) (1979), 67-77. [2] B. G. Hayes, W.M .Michael and D. C. Darren, Implants for cranioplasty, Otolaryngologic Clinnics of North America 28 (2) (1995), 381-400. [3] P.S. D' Urso, W.J. Earwaker, T.M. Barker, M.J. Redmond, R.G. Thompson, D.J. Effeney and F.H. Tomlinson, Custom cranioplasty using stereolithography and acrylic, British Journal of Plastic Surgery 53 (2000), 200-204. [4] K.Remsen, W. Lawson and H.F. Biller, Acrylic frontal cranioplasty, Head and Neck Surgery 9 (1986), 32-41. [5] J.M.Joffe, S.R.Nicall, R. Richards, A.D.Limey and M.Harris, Vailidation of computer assisted manufacture of titanium plate for cranioplasty, Int. J.Oral Maxillofac. Surg. 28 (1999), 309-313. [6] M. Perry, P.Banks, R. Richards, E.P.Rriedman, and P.Haw, The use of computer-generated three-dimensional models in orbital reconstruction, British journal of Oral and Maxillafacial Surgery 36 (1998), 275-284. [7] H. Eufinger, M. Wehmoller, E. Machtens, L. Heuser, A. Harders and D. Kruse, Reconstruction of craniofacial bone defects with individual alloplastics implants based on CAD/CAM-manipulated CT- data, J.Cranio Maxillo-facial Surgery 23 (1995), 175 –181. [8] H. Eufinger, M. Wehmoller, A. Harders, and L. Heuser, Prefabricated prostheses for the reconstruction of skull defects, Int.J.Oral Maxillofac. Surg. 24 (1995), 104 –110. [9] H. Eufinger, A.R.M. Wittkampf, M.Wehmoller and F.W.Zonneveld, Single-step fronto-orbital resection and corresponding titanium implant: a new method of computer-aided surgery, J. Craniomaxillofac Surg. 26 (6) (1998), 373-378. [10] H. Eufinger, and M.Wehmoller, Individual prefabricated titanium implants in reconstructive craniofacial surgery: clinical and technical aspects of the first 22 cases, Plastic and reconstructive surgery 102 (2) (1998), 300-308. [1]


[11]

[12]

[13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

11

P.S. D’Urso, R.L. Atkinson, M.W. Lanigan, W.J, Earwaker, I.J. Bruce, A Holmes, T.M. Banker, D.J. Effeney and R.G. Thompson, Stereolithographic (SL) biomodelling in craniofacial surgery, British Journal of Plastic Surgery 51 (1998), 522-530. I. Onno, H. Gunji, F. Kaneko, S. Nu,azawa, N. Kodama and S. Yoza, Treatment of extensive Cranial bone defects using Computer-Designed Hydroxyapatite Ceramics and Perosteal Flaps, Plastic and Reconstructive Surgery 92 (5) (1993), 819830. B.Tomancok B, G. Wurm, J.Trenkler and Manfred, Extended cranioplasty with prefabricated carbon fiber reinforced plastic implant, Phidias - EC funded Network Project on Rapid Prototyping in Medicine 3 (1999), 7-9. B.A. Toth, D.S.Ellis and W.B.Stewart, Computer-designed Prostheses for Orbitocranial Reconstruction, Plastic and Reconstructive Surgery (1988), 315-324. H.M. Connell, P.F.Statham, D.A.Collie, F.S. Walker and K.F.Moos, Use of a template for custom cranioplasty. PhidiasRapid Prototyping in Medicine 2 (1999), 7-8. M. Fujimoto and K. Kariya, An improved method for digitized data reduction using an angle parameter, Measurement 12 (1993), 113-122. J.Vander Sloten, K.Degryse, R.Gobin, G.Van der Perre, and M.Y.Mommaerts, Interactive simulation of cranial surgery in a computer aided design environment, Journal of Cranio-maxillofacial surgery 24 (1996), 122-129. J.T. Lambrecht, and F.Brix, Individual skull model fabrication for craniofacial surgery, Cleft Palate J. 20 (1990), 382-386. G. Santler, H. Karcher and C.Ruda, Indications and limitations of three-dimensional models in cranio-maxillofacial surgery, Journal of Cranio-Maxillofacial Surgery 26 (1998), 11-16. J.S. Bill, J.F. Reuther, W.Dittmann, N.Kubler, J.L.Meier, H.Pistner and G.Wittenberg, Stereolithography in oral and maxillofacial operation planning, Int.J.Oral Maxillofac. Surg. 24 (1995), 98-103. Y.Y. Yau, J.F. Arvier, T. M. Barker, Technical note: Maxillofacial biomodeling – preliminary result, The British Journal of Radiology 68 (1995), 519-523. A. Pompili, F. Caroli, L.Carpanese, M. Caterino, L. Raus, G. Sestili and E. Occhipinti, Cranioplasty performed with a new osteoconductive, osteoinducing hydroxyapatite-drived material, J.Neurosurg. 89 (1998), 236-242. G. Santler, H. Karcher, A. Gaggl, and R. Kern, Stereolithography versus milled three-dimensional models: comparision of production method, indication, and accuracy, Computer Aided Surgery 3 (1998), 248-256.