From DICOM to the Operating Table: How a Veterinary CT Scan Becomes a 3D-Printed Bone Model

A surgeon who can hold the fracture in her hands before she ever picks up a scalpel makes faster, better decisions in the OR. That single idea is why 3D-printed bone models have moved from novelty to standard-of-care planning tool in advanced veterinary practice. But the model is only as good as the data and the workflow behind it. Here is exactly how a veterinary CT scan becomes an anatomically accurate, sterilizable bone model you can plan on.

It Starts With the DICOM, Not the Printer

Every medical and veterinary CT scanner exports its slices in a format called DICOM (Digital Imaging and Communications in Medicine). A DICOM dataset is a stack of cross-sectional images plus the metadata that tells software how thick each slice is and how far apart they sit. The quality of everything downstream is decided here, at acquisition.

Three settings matter most for a printable model:

• Slice thickness. Aim for 0.625 mm to 1.0 mm. Thick 3 mm slices, common in routine diagnostic scans, produce a stair-stepped, blurred surface that no amount of post-processing fully recovers.
• Bone (high-frequency) reconstruction kernel. Soft-tissue kernels smooth away exactly the cortical detail we need.
• Minimal motion and metal artifact. A sedated patient and artifact-reduction protocols keep the edges crisp.

Step 1 — Segmentation: Separating Bone From Everything Else

Segmentation is the process of telling the software, “this voxel is bone, that one is not.” Because cortical bone is dense, it shows up bright on CT, so we start with a threshold—a brightness cutoff measured in Hounsfield Units. Everything above the cutoff becomes part of the model.

Thresholding alone is never enough. Young animals with thin cortices, osteopenic bone, and joint spaces where two bones nearly touch all require manual cleanup. A technician separates the femur from the tibia at the stifle, removes the scan table, and closes any false gaps. This is the step where anatomical knowledge matters as much as software skill, and it is where a rushed service produces a model that looks right but reads wrong on the table.

Step 2 — From Voxels to a Watertight Mesh

Once bone is isolated, the software converts the segmented volume into a surface mesh—a shell of tens of thousands of triangles, exported as an STL file. A raw mesh is rough: it has terraces from the slice spacing and a faceted surface. Cleanup involves:

• Smoothing just enough to remove staircase artifacts without erasing real anatomy.
• Wrapping and hole-filling so the mesh is “watertight”—a single closed surface a printer can interpret.
• Verifying wall thickness on thin structures like the orbital wall or a young patient’s cortex so they survive printing and handling.

If you have ever wondered why two services quote wildly different prices for “the same model,” this is usually why. A clean, validated mesh is hours of skilled work; a one-click auto-export is minutes. They are not the same product. (We cover the broader file-prep principles in our guide on preparing files for 3D printing.)

Step 3 — Choosing the Right Print Process

Not every print technology suits a surgical model. The choice depends on what the surgeon needs to do with it.

For most orthopedic planning—TPLO, complex fracture reduction, angular limb deformity correction—a tough nylon model that tolerates drilling and a steam-autoclave cycle is the workhorse. If you need help matching the job to a process, our which-process guide walks through the trade-offs.

Why It Pays Off in the OR

The clinical case for pre-surgical models is no longer anecdotal. Holding the actual geometry lets a surgeon select implant sizes ahead of time, contour plates on the bench instead of in the patient, and rehearse the cut sequence. The result is shorter anesthesia time, less intraoperative improvisation, and a concrete tool for explaining the procedure to an owner. We walk through a focused example in our article on 3D-printed canine TPLO models.

What We Need From You to Get Started

The handoff is simpler than most clinics expect:

1. The full DICOM dataset (not exported JPEGs or a PDF report)—thin-slice bone protocol if possible.
2. The anatomy and laterality of interest, and the surgical question you are trying to answer.
3. Your timeline. Surgical cases can be expedited.

From there we handle segmentation, mesh validation, process selection, and printing as a licensed engineering shop—the same rigor we bring to production manufacturing, applied to anatomy.

PartSnap is a licensed professional engineering firm offering 3D printing, CAD, and manufacturing services. Veterinary models are surgical planning aids and do not replace clinical judgment.