The authors' objective in this project was to replace current state-of-the-art manual methods for preoperative production (i.e., prefabrication) of large-format (>100 cm2) cranioplasties with a system for computer-aided design and direct computer-aided manufacture of the implant's shape. This system uses standard 3D CT data, requires no specialized training, and produces an accurately fitting cranioplasty that can be recast in the physician's material of choice (e.g., polymethylmethacrylate [PMMA] or pre-bent titanium plating). The authors begin by locating the cranial defect margin on a skull surface image generated from a 3D head CT-scan. A right-to-left mirrored or average 3D skull surface template image is then fit to the patient's skull surface image. The area around the defect is cut out and stitched to the previously isolated defect margin. This defect-filling surface is then tapered and 3D printed. The 3D printed implant model is then recast in a biocompatible material. Manually generated cranial implants produced for five patients were compared with implants resulting from this new computer-based method. All five computer-generated implants were better fitting and more cosmetically suitable than the manually generated skull plates received by these patients. These well-fitting implants are more likely to protect the brain from trauma and infection. Therefore, the authors conclude that their new production method provides a better result with less expense than current methods for preoperative or intraoperative fabrication of large-format cranioplasties.
Large-format skull defects are not uncommon sequelae to trauma, cancer, stroke, and reconstructive surgery. Intraoperative repair is time-consuming and often results in suboptimal cosmesis and insufficient protection from trauma and infection. Recently, three-dimensional (3D) CT data has been used to render full-size, anatomically accurate models of patients' skulls. 1,2 The patient's skull anatomy is reproduced via 3D printing for presurgical use in manual implant design and production. While several workers have noted that custom prefabrication of large-format cranial implants (e.g., 100 cm2 or greater surface area) reduces risk to the patient (e.g., better fit, better cosmesis, less invasive surgery, reduced surgical duration, faster recuperation) and reduces cost (i.e., via reduced intraoperative time), craniofacial surgeons do not routinely recommend prefabricated implants to their patients with large cranial defects. 1,3,4
We expect that there are at least two reasons why this less costly, therapeutic technology is seldom used. First, surgeons are most comfortable repairing boney defects with autograft material due to the good chance that it will be successfully incorporated by defect site tissues. However, problems of donor site morbidity, sufficient donor material, and crafting the correct shape from available donor material increase dramatically with the large-format defects that are the subject of this study.
Second, the prosthetics industry's model for craniofacial and, especially, orthopedic implants is to create “ideal” shapes in a range of sizes. Such ready-made implants are easier to distribute. However, hospitals now commonly penalize this model in favor of strategies, such as patient-specific prefabrication, that save intraoperative time. Many hospitals no longer pay to stock prostheses, instead opting to reimburse vendors only after implantation. Some of the materials used for off-the-shelf implants allow for some intraoperative modification to improve fit. However, it is often difficult to obtain good host–implant adjacency intraoperatively with gloved hands, scalpels, and air drills. Additionally, if an implant fails during modification or cannot be adapted to the patient, it is billed but not reimbursed.
Appropriate fit and function for a given patient are very difficult to anticipate with off-the-shelf prostheses. Therefore, in many cases the patient's anatomy must be substantially modified to fit the implant. Often the unmodifiable structural aspects of the implant are designed to bring about controlled stress-strain relationships. The design of most implants attempts to control strain-stress relationships to those that minimize implant failure. However, with the rapidly increasing range of biocompatible materials, especially polymers, there is need for a better understanding of the stress-strain relationships in each patient. Another new priority is to heal as much of the defect site as possible (i.e., tissue engineering), thereby facilitating a return to the site's original function. Prefabricated implants designed from high-resolution images are more likely to accurately fit the patient. The chance to design a patient-specific implant also opens the possibility of designs that improve biomechanical outcomes and promote the normal bony healing response in tissue engineered grafts. 5,6
Obtaining a custom-fitting, prefabricated cranial implant is not yet straightforward. US vendors currently offering customized prefabricated cranial implants (e.g., Porex [Newman, GA], Medtronic [Minneapolis, MN], W. Lorenz [Jacksonville, FL]) guide the surgeon to provide them with a 3D CT-scan of the patient's cranium. Most craniofacial surgeons and their radiologist colleagues are not in the habit of acquiring CT data for model generation. These scans require uncommonly high resolution, fiduciary devices to provide independent verification of scan accuracy, and artefact reduction control. In most cases the referring surgeon and radiologist have no means to prevent, nor devices to record the lack of, patient movement during the scan. Finally, the surgeon and radiologist often have difficulty exporting CT data from the CT console, especially to an archival media and format that will be intelligible by the vendor's software. Most hospitals in the United States have not greatly reduced or eliminated their use of film and have only begun to adopt digital radiological data-handling protocols and equipment. Additionally, the issues of accurate 3D CT volume data acquisition are a significant impediment to widespread utilization of 3D CT as source data for large-format custom cranial implant manufacture. Fortunately, craniofacial surgeons are sensitive to all of these issues as they commonly use CT imagery and image-guided surgical systems.
If the surgeon and radiologist are comfortable with these methods, they may find they cannot use a custom, large-format, cranial prosthetic because production costs are too high for reimbursement. Costs are high because current commercial cranial prosthetic prefabrication methodologies require three expensive and time-consuming steps to produce a prefabricated large-format cranial implant from the patient's 3D CT-scan. The first step is the painstaking manual identification of the patient's skull, which allows removal of the rest of the items in the scan (i.e., skull segmentation). This process is often partially automated by simple thresholding tools. However, a simple thresholding approach may result in a poorly fitting implant due to inaccurate 3D CT skull representation of areas with foraminae, thin surfaces, or CT slice-edge (i.e., stairstep) artifacts. The second step requires printing much of the patient's skull, usually all of the cranium. The national vendors (mentioned above) who prefabricate cranial implants print a patient's skull model segmented from the 3D CT-scan data via 3D Systems (Valencia, CA) SLA stereolithographic device or a similar rapid prototyping 3D printing device. Stereolithography provides 0.1-mm resolution, sufficient accuracy for a model to inform production of a large-format cranioplasty; however, the cost for highly accurate SLA skull models ranges from US$1000–5000, depending on the size. One cannot limit the 3D printing to the region around the defect alone as the third step in current production strategies requires that an anaplastologist or prosthodontist use the available nondefective cranial morphology to manually design an implant. This process relies on visual inspection of the potential implant's seating within the skull defect. This manual interaction with the model requires hand skills sufficient to produce an implant that provides cranial symmetry, a good fit, and, most importantly, sufficient strength to protect the underlying brain from trauma and infection.
We hypothesize that the cost of a prefabricated cranioplasty would be halved, and that utilization of this technology would surge, were we to translate the manual implant design procedure to the computer. Computer-aided design (CAD) would allow direct SLA production of implants. This in turn would (1) save the time and cost of production of skull models, (2) eliminate the need for an anaplast or prosthodont to use that skull model to manually produce an implant, and (3) might allow direct printing of the implant in an implantable material 5,6 We present a rapid prototyping method for the computer-aided design (CAD) and computer-aided manufacture (CAM) of five large-format cranial implants. The quality of the resulting implants is compared with the implants these patients received. We wish to determine whether these computer-generated implants would be as useful, or more so, than manually produced implants which are informed by SLA models of the patient's skull.
*Department of Neurological Surgery and the Research Institute, University Hospitals of Cleveland, and Department of Neurological Surgery, Case Western Reserve University, Cleveland, Ohio, †(at the time of this project) Department of Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, Ohio, ‡(currently at) Osteoplastics Corporation, Solon, Ohio
Address correspondence to David Dean, Department of Neurological Surgery, Hanna House 5th Floor, M.S.# 5042, University Hospitals of Cleveland, 11100 Euclid Avenue, Cleveland, OH 44106 USA; e-mail: DavidDean@cwru.edu
Sources of Support: Partial support was derived from NIH SBIR Grant# R43-DE013786, “Image-guided reconstruction of neurocranial deficit.” Partial support was provided by the Research Foundation, Department of Neurological Surgery, UHC/CWRU.
Disclosure Statement: Data obtained from human subjects was done under an approved University Hospitals of Cleveland Internal Review Board human subjects protocol. The first author is a co-founder of, consultant to, and shareholder in Osteoplastics Corp., Cleveland, OH. Osteoplastics is the recipient of the cited SBIR grant that partially supported the study reported here. Osteoplastics provided this support to the first author's Department at Case Western Reserve University via an NIH consortial subcontract.
This work was first presented at Computer Assisted Surgery 2001, October 12 in Nürnberg, Germany.