Fig. 2 3D printed model of a large sacral tumor shown exiting the obturator foramen; used to help in surgical planning.
Courtesy of Peter Rose, MD, Mayo Clinic, Rochester, Minn.


Published 4/1/2015
Hani Haider, PhD; Karthikeyan E. Ponnusamy, MD; Nicholas John Giori,MD; Paul A. Anderson, MD; Ahmad Nassr, MD

One Layer at a Time: Rapid Prototyping in Orthopaedics

What’s the state of the art in orthopaedic 3D printing?

Three-dimensional (3D) printing is a popular umbrella term to describe additive manufacturing, a form of rapid-prototyping. It refers to building an object, one thin layer at a time.

In the 1980s, stereolithography used a fine laser to solidify layers of liquid thermoplastic resin. A decade later, inkjet-printer-like heads used a liquid binder to solidify powder layers. Today, many other processes can be used. (See sidebar “Additive manufacturing processes.”) The thinner the layers, the finer the laser, binder, or material extrusion nozzle, and the more accurate the nozzle motion control is, the higher the resolution of the final product, which is now reaching less than 0.002 inches.

Unlike subtractive manufacturing (such as milling and turning in which material is removed from solid blocks or rods until the desired object is fabricated), additive manufacturing allows highly complex 3D objects to be built. The current boom in additive manufacturing technology is rapidly improving its resolution, speed, and economy. For orthopaedics, this technology will have a major impact on the next generation of healthcare devices, such as anatomic models, polymer and metal devices and instruments, and tissue-engineered implants.

This article describes the technology behind additive manufacturing, its various techniques and materials, and the use of models for preoperative planning and resident education.

The technology
Custom replica models of a patient’s anatomy can be manufactured based on 3D Digital Imaging and Communications in Medicine (DICOM) format data from CT or MRI images. A DICOM dataset typically includes a large number of related two-dimensional (2D) image files. Each file/image contains patient identification data, imaging study settings, and slice/image parameters (all called “headers”), as well as X/Y numerical coordinate data for each point in the image.

Each data point also has an associated Hounsfield unit (HU), a quantitative scale that is related to the density of tissue at that point. Together, the 2D images (supplemented with Z values, signified by the image sequence number/position) indirectly create a 3D matrix of data from which tissue type may be later inferred (as a level of whiteness/brightness on 2D films).

The DICOM data then undergo a series of software processes. The first is “segmentation” in which the data are classified, based on the HU values, as possibly belonging to certain tissue types or elements. This segmentation may also be applied to selected (geographic) parts of the data. In orthopaedics, for example, a DICOM dataset may be segmented to identify all points in the 3D domain that have HU above a certain threshold as solid bone. Regions within the dataset may also be specified to segment out only a certain bone or parts of it.

The next software process is “reconstruction,” which generates a “surface” or “volumetric” model. A surface model is more common and efficient and can be represented in a data file as a stereolithography (STL) format. An STL surface model contains only the points on the surface in a triangulated mesh that represents the 3D shape of the object, but is essentially hollow. Such a surface model may be treated as a “skin” around the object that reflects light and shading for computer visualization. The higher the resolution of the CT images, the smaller the surface triangles, resulting in a smoother, more realistic object. However, this adds to the file size and requires considerably more computer memory and processing power.

In post-processing, mathematical functions can be applied to smooth excessively jagged edges between image layers, or within a layer. Such smoothing can change the shape of some triangles, make them smaller, and/or add extra-small triangles, also adding to the size of the STL file.

An STL format enables efficient graphic rendering of the surface of a shape on a computer. It also requires less material quantity to physically construct (eg, on a 3D printer) than a true solid (not hollow) model. Attaching multiple surface models to each other can enable the construction of any 3D object. The result is fidelity in visual representation, efficiency of computational processing, economy, and speed for rapid prototyping to manufacture models.

Many commercially available specialist software applications are available for importing and visualizing raw DICOM data, performing the segmentation and 3D reconstruction, and producing STL files among other suitable formats. Literally all additive manufacturing process machines (such as 3D printers) can import and manipulate STL (or other surface model files) to build the physical 3D objects.

Diagnostic and surgical planning models
Medical models built in this way can help both patients and physicians understand complex patient-specific anatomy encountered in pelvis and spinal deformity cases. The models can be used preoperatively or intraoperatively and for education or research.

Fig. 1 (A–C) shows the value of this approach. The radiograph (Fig. 1A) shows a patient with a complex hip problem: a functioning hip replacement on one side and a plate repair on the other side, which also needs a hip replacement. Comprehensive 3D planning was used to improve the quality of fixation and to avoid potential leg length discrepancy. Fig. 1B shows the STL files of the bones and plate hardware produced after segmentation and reconstruction. The grey plate was segmented separately from cortical bone, based on its higher metal density.

Fig. 1C shows the physical models manufactured by an early STL machine. As a result of this preoperative planning and the surgeon’s skill, the patient had a successful hip replacement with good fixation and no leg length discrepancy.

Models can also be used preoperatively to help patients understand their condition and the surgery. They can also be used to simulate trial osteotomies or other procedures. Softer structures such as nerves, blood vessels, or tumors (Fig. 2) can be depicted in different colors to help surgeons develop surgical plans. The same physical models can be used intraoperatively to help guide the surgeon, along with any other imaging modalities.

One study found that using fabricated models to guide fluoroscopy during surgery for atlantoaxial instability resulted in reduced surgical times. Models made from sterilizable materials can be placed in the surgical field to help with precontouring of plates or rods. A negative mold of a defect can be used intraoperatively to size and shape grafts. Models for preoperative planning and intraoperative guidance could help surgeons become more accurate and faster and may enable more minimally invasive approaches.

Models can also be used to educate residents and enable them to practice osteotomies and other techniques. The models can even simulate different pathologic bone qualities. Pathologic bone models can be used in implant design and testing. Tissue-engineered models could be used in pharmaceutical research.

Although the possibilities of this new technology are endless, they are bound to be associated with challenges and risks. Next month, a follow-up article will cover additional applications and challenges for 3D printing and provide a look into the future of this technology.

Hani Haider, PhD, Karthikeyan E. Ponnusamy, MD, Paul A. Anderson, MD, Nicholas John Giori,MD; and Ahmad Nassr, MD, are members of the AAOS Biomedical Engineering Committee. One or more of the authors report potential conflicts of interest; for more information, visit

Additive Manufacturing Processes
Special photopolymers are used with lasers in stereolithography and Polyjet™ processes, in which ultraviolet light-curable materials are extruded through an inkjet head one layer at a time and then cured.

The fused deposition modeling method deposits plastic wire or thin ribbon from a supply reel in a sequence of layers to form 3D objects. This method uses common plastics such as acrylonitrile butadiene styrene, nylon, polycarbonate, polypropylene, and polyether ether ketone (PEEK).

Selective laser sintering (SLS) is a more refined process that uses fine lasers to melt and solidify thin layers of thermoplastic nylon powder or polymethyl methacrylate (PMMA).

Laminated object manufacturing can be considered additive because thin sheets of any material are cut in advance into different two-dimensional shapes and then assembled to adhere together into the desired 3D shape with any internal cavities.

For orthopaedic surgery, the most exciting additive manufacturing process may be direct metal laser sintering (DMLS). This process fuses layers of atomized metal alloy powder to form objects. Stainless steel, titanium, cobalt-chromium, aluminum, and nickel-chromium may be sintered at various densities. With DMLS, manufacturers can create durable instruments and implants that can be sterilized and reused. DMLS is the most expensive process, however, and although used by many manufacturers, it has not yet widely entered academic or small company research and development labs.