Polygonica is an ideal toolkit for developing surgical planning and orthotic design applications. The SDK provides a wide range of algorithms for working directly with raw scan data without the need for complex, time consuming and expensive reverse-engineering operations. But, if reverse-engineering is absolutely necessary, Polygonica also provides a range of mesh preparation tools that provide faster and easier segmentation for BREP-surface creation.
In this blog we’ll explore two common medical workflows focused on creating custom implants, based on meshes created from a patient’s CT scan.
- Orthognathic surgery planning
- Craniomaxillofacial surgery planning
As always, the blog is supported by accompanying video demonstrations, and the code for the reference app is available to Polygonica customers and evaluees on an unsupported basis. Please feel free to get in touch via the contact page.
Orthognathic Surgery Planning
Medicine loves to use long complex words which baffle and bemuse. And I love a digression.
According to Google, ‘ortho’ is from the Greek for straight, and ‘gnathos’ is from Greek for ‘lower jaw’. At least that’s the simple explanation. I suspect the real derivation is even more complex.
So orthognathic surgery covers all types of surgery designed to correct misalignment in the jaw region. One specific variant is malocclusion, or poor bite alignment, which is more commonly corrected using braces and aligners. In more severe cases surgery may be necessary.
Mal-occlusion derives from the Latin for bad or incorrect, and occlusion is from the Latin for ‘shut up, close up’. At some point in the middle of the 19th Century occlusion was adopted by the dental profession to specifically refer to how the teeth in the maxilla (Latin, ‘upper jaw’) and teeth in the mandible (also Latin, ‘jawbone’, ‘used for chewing’) meet.
So we could just say upper jaw, lower jaw and bad bite, but it just doesn’t sound as cool.
End of digression.
Orthognathic surgery generally involves cutting the skull and jaw such that the bones can be moved to ‘straighten’ the alignment, then anchoring the adjusted bones in place using metal plates.
In many cases the metal plates are standard pre-manufactured parts, which the surgeon then bends so that they fit into place.
With the advent of scanning and 3D printing, custom parts can now be manufactured. In this blog we’ll cover using Polygonica in both workflows.
Cutting the skull and jaw
In order to adjust the misaligned parts of the skull and jaw, the surgeon needs to cut through the bones so that the parts can be moved around with respect to each other. The implants will then stabilize the parts in the new position.
As shown in the videos at the end of the blog, Polygonica’s mesh Boolean is ideal for modelling these cutting steps in surgical planning software.
The split parts can then be manually selected and transformed into the appropriate position.
Using shrinkwrap to create a smooth surface across gaps
After the surgeon has realigned the bones, there might be significant differences in the bone heights on either side of a gap that needs to be secured using a printed implant.
Whilst the profiles required to smoothly bridge such large gaps could be manually specified by the surgeon during planning, this is inefficient. We want to find a way to automatically bridge such gaps without creating unnecessary large concavities in the implant strips.
Polygonica’s shrinkwrap helps solve this problem by providing a tight and accurate wrap around the surface of the bone scan, and a smooth, almost convex, bridge across the large gaps. An example based on the complex ankle joint of an animal is shown below.
Before creating the implants a shrinkwrap of the entire skull is generated to provide smooth surfaces across the cuts made by the surgeon.
Creating a custom 3D printed implant direct from the scan
The workflow for generating the custom implant for the strips is quite similar to generating dental aligners, guides and crowns, which we’ve covered in previous blogs:
For orthognathic strips the following approach is taken:
- Identify the locations on the scan for the implant, in this case by the surgeon selecting anchor points directly on the skull model
- The selected points are joined using Polygonica’s geodesic path finding, and then Polygonica’s radial curve offset is used to inflate the curve across the surface of the shrinkwrapped version of the model to create the outer boundary loops. Using the shrinkwrap ensures the resultant strip smoothly bridges more significant gaps.
- New edges are imprinted into the mesh along the offset curves so that the faces that lie on the surface inside the offset curve can be retrieved. This is where the implant will lie on the surface – or it can be adjusted so it lies slightly above it.
- These faces form a continuous sheet, which is then thickened using Polygonica’s spherical 3D offset to create the implant itself. In practice the cylindrical strip anchors are created separately from the strips that join them, and then the results are joined together using Polygonica’s mesh Boolean.
- Remesh the strip to improve the quality of the faces in the mesh that will be printed
- Smooth the outer surfaces of the mesh.
- Boolean subtract cylindrical holes where screws will be inserted.
Placing stock implants using Polygonica
Not all implant workflows use custom 3D printed implants. In many cases surgeons use off-the-shelf implants which are bent into the required shape before being deformed further whilst being screwed into place on the skull.
In this case the implants are not manufactured from the custom meshes created using Polygonica, but a Polygonica-based planning workflow can help the surgeon decide if the parts are the correct size and can be adjusted to the required fit.
Polygonica’s deformation algorithms can be used to simulate bending the strips along the creases provided by the manufacturer.
However, in the reference app we’ve chosen to use a uv mapping of the profile of the implant strip. This has the advantage that the placement will be visually correct even in regions of high curvature. The imprinted curve can then be moved interactively across the surface to make small adjustments, with a re-imprint step at the end to ensure the mapping of the final location is accurate.
As before, once the profile is imprinted, a model of the strip can be recreated from the mesh surface, showing how the implant is likely to deform when the surgeon screws it into the bone.
Craniofacial, maxillofacial and craniomaxillofacial surgery planning
We’ve already learned from our casual etymology that maxillo refers to the upper jaw. Unsurprisingly facial means just that – ‘related to the face’. Cranial is from the Latin for skull.
So in these workflows we’d expect to be planning surgery anywhere from the upper jaw to the top of the skull. However maxilla is also used as a term to refer to both upper and lower jaws, so craniomaxillofacial surgery seems to cover pretty much anything bone-related and above the neck.
But if I get lots of letters from surgeons correcting me, I’ll be very happy.
Creating a large skull cap implant
In the case we’re demonstrating a large area of bone is missing from one side of the skull. We want the implant to completely cover this area.
Using Registration and Alignment
Our first challenge is to define a surface onto which we can create the implant geometry. As the hole is on one side of the skull, we can take advantage of the fact that skulls are generally symmetrical.
So, first a simple mirror transform is applied to a copy of the skull. This then needs to be aligned with the original. Since the result of the mirror transform and subsequent translation to match origins is heavily dependent on the mesh’s coordinate origin and its relation to the best plane of symmetry, we found it more reliable to use Polygonica’s advanced mesh registration/alignment API, PFSolidRegister. This gives us a close fit of the mirrored model to the original, such that the mirrored surface can be used as the surface from which to create the skull cap.
Should the model have significant asymmetry then, following a rough registration, the function PFTransformMatrixCalcFromPoints can be used to select equivalent points on both the original mesh and the mirrored mesh, after which fine registration can be re-run to improve the final fit.
Creating and customizing the implant
This gives a rough surface on which to create the skull cap. From here we follow a similar workflow to before:
- Manually select points on the mirrored surface around the location of the hole in the original surface
- Create and smooth an imprint curve on the copy surface
- Emboss edges around the curve
- Select faces inside the curve
- Offset (sheet-thicken) these faces to create the skull cap
- Blend (fillet) the sharp edges
- Remesh and smooth top surface of the plate
That gives us our basic implant. Now for the bells and whistles:
- Manually select and Boolean subtract flexibility cut lines
- Manually place and engrave text using text and label embossing / debossing
- Boolean subtract perforation holes across the top surface of the plate
- Manually place and Boolean subtract screw holes with conical counter-sinking
And voila, a custom-designed implant ready to pass to your favourite print preparation software (which might also be using Polygonica).
You can watch live demos of the orthognathic print workflow and the cranial implant workflow on the Polygonica YouTube channel via the links below.
