3D printing makes hospital surgical planning easier

A surgical model starts as a DICOM series, usually a CT scan with 0.5–1.0 mm slice spacing. Thin slices matter. At 2 mm spacing a 1 mm cortical shell can disappear between slices, and the segmentation software will fill the gap with something that was never in the patient. For bone work, 0.625 mm to 1.0 mm is the working range most radiology departments can provide.

Segmentation is where the real labor sits. Bone thresholds are straightforward because the Hounsfield values are high and the boundary is sharp. Soft tissue, vessels and nerve bundles need either contrast enhancement or manual editing, which can take hours per case. This is the step that decides whether the model is clinically useful or just a rough shape.

After segmentation comes mesh cleanup and a format conversion to STL or a similar mesh file. Holes are filled, floating artifacts are removed, and the mesh is decimated to a workable triangle count. Non-manifold edges will stop most slicers. A watertight mesh, checked in software rather than by eye, saves a failed print and a lost day.

Scale is the last check before printing. A model exported in inches and sliced in millimeters comes out 25.4 times too large. It happens more often than people admit, and it is why a printed ruler or a known landmark dimension belongs on the build plate next to the model.

Material jetting and PolyJet-style printers dominate anatomical models. They hold thin walls, print in multiple colors, and can show a tumor in one color and surrounding bone in another. A spine model with a colored nerve root is easier to read in a room than a single-color part. The trade-off is cost per part and the need to remove support material from fine features.

Stereolithography (SLA) gives the smoothest surface of the resin processes and is the common choice for cutting guides and drill templates that must sit flush on bone. Dimensional accuracy is good, but the resin must be handled and cleaned properly, and thin guide walls can warp during post-cure if the geometry is not supported.

Fused deposition modeling (FDM) is the low-cost option. Layer lines are visible, fine detail is limited, and it is best used for large, simple teaching models or for checking fit before committing to a resin part. For a patient-specific guide that contacts bone, FDM is usually not the right process.

Selective laser sintering (SLS) in nylon produces tough, autoclavable parts with good mechanical strength. It is a reasonable fit for handles, brackets and larger anatomical models where surface finish is secondary. Titanium and cobalt-chrome laser melting is a different category: it is used for permanent implants and for metal cutting guides, and it brings a much longer qualification path.

The clearest wins are in complex bone anatomy where a 2D screen flattens depth. Pelvic and acetabular fractures, spinal deformity correction, and mandibular or midface reconstruction are the standard examples. In these cases the surgeon has to judge angulation and clearance in three dimensions, and a physical model shows it in seconds rather than minutes of rotating a viewer.

Congenital heart defects are another strong fit. A printed model of a small, malformed ventricle lets the team plan a patch shape and a cannulation site before opening the chest. The model is also useful in the conversation with the family, because a printed heart is easier to explain than a rendered image.

Orthognathic and craniofacial surgery benefits from guides rather than models alone. A printed splint or osteotomy guide transfers the digital plan to the operating field. The accuracy of the transfer depends on how well the guide seats on the exposed bone, which is a design question more than a printing question.

Where printed models add little: routine appendectomy, standard total knee replacement with an off-the-shelf system, and any case where the imaging already gives an unambiguous answer. Printing takes time and money. If the plan will not change, the model is a teaching aid, not a planning tool.

• 1
  Good fitPelvic and acetabular fractures, spinal deformity, mandible reconstruction
• 2
  Good fitCongenital heart defects, complex craniofacial cases
• 3
  Weak fitRoutine procedures with an unambiguous imaging answer
• 4
  Deciding factorWill the printed part change what the surgeon does?

Model accuracy is limited by the scan, not by the printer. A printer that holds ±0.1 mm on a build is accurate enough. The scan slice spacing, the segmentation threshold, and the smoothing applied during mesh cleanup together contribute more error than the printing step. Chasing a tighter printer tolerance while segmenting at 2 mm is wasted effort.

Segmentation error is systematic, not random. A threshold set slightly high shrinks the bone surface by a fraction of a millimeter across the whole model. That is fine for a teaching model and not fine for a guide that must seat on a prepared surface. The fix is a known reference: a fiducial marker, a scanned calibration phantom, or a landmark measured on the scan and checked on the printed part.

For printed guides that contact bone, the seating surface is the critical feature. A guide that rocks by 0.3 mm will place a drill hole off target. Design the guide with a wide, unambiguous seating area and a clear insertion path. Narrow contact patches and undercuts are the usual causes of a guide that does not sit flat.

For metal parts, machining tolerances are a different world. When a surgical instrument or a trial component is machined rather than printed, we work to ±0.005 mm on critical features with surface finish between Ra 0.2 μm and Ra 1.6 μm depending on the function. Printed polymer models sit in a looser band, and mixing the two tolerances in one specification is a common mistake.

A model that only touches a table can be any clean resin. A guide that enters the sterile field needs a material and a process that survive the chosen sterilization method. Steam autoclave at 134 °C rules out most photopolymer resins. Where steam is required, nylon SLS or a metal part is the safer route. Where the guide stays outside the sterile field, resin is fine.

Biocompatibility is a separate question from sterilization. A material that survives an autoclave cycle is not automatically suitable for prolonged tissue contact. For anything that touches the patient, the material supplier's documentation has to match the intended contact duration and the regulatory path of the hospital. We do not substitute one resin for another on a medical job without that check.

Documentation is what separates a hospital project from a general prototyping job. The print file version, the material lot, the machine, the build orientation and the post-processing steps should all be recorded. If a guide is later found to be off, the record is what allows the team to trace the cause. Our quality system is certified to ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022, and we hold 100% inspection before shipment with reports on request.

Patient data needs its own controls. DICOM files carry identifiers. An NDA and a defined data-handling path are standard on medical work, and uploads are kept secure and confidential. De-identified datasets with a coded case number are the practical default for print work.

A printed anatomical model from a clean DICOM set can move quickly once segmentation is agreed. Quotation and a free DFM analysis come back within 12 hours, production can start within 24 hours, and parts ship in 3–5 days. For a scheduled case, that window is usually enough. For an urgent case, it often is not, and the team should know that before the request goes out.

The bottleneck is rarely the printer. It is the segmentation review, the surgeon's confirmation that the model matches the imaging, and the reprint if the first version is wrong. Build a review step into the schedule. One extra day of review is cheaper than a reprint on the morning of surgery.

When a project involves both printed models and machined metal components, keeping them in one workflow saves a handoff. We run 5-axis, 4-axis and 3-axis machining, mill-turn, 3D printing, vacuum casting, die casting and sheet metal under one roof, with no minimum order quantity from a single prototype upward. That matters when a case needs one printed guide and one machined trial instrument.

The practical rule: start the request as early as the imaging exists, not after the surgical date is fixed. Early requests leave room for a reprint, and a reprint is the cheapest insurance in this workflow.