Patient-specific 3D implants

Lisa Ernst · 21.11.2025 · Health · 10 min

When a surgeon from a Swiss hospital calls us because a piece of skull is missing after an accident and the surgery is planned for two weeks from now, the clock immediately starts ticking in our workshop. In such situations, you quickly realize whether your workflow for patient-specific 3D implants really works or if everyone is still chaotically sending emails and DICOM files back and forth. Within a few days, a clean 3D model is created from a CT dataset, and from that, an implant or an anatomical model that fits this one patient exactly. In practice, we repeatedly experience how much more relaxed a team goes into the OR when they have literally had the anatomy in their hands beforehand.

This is precisely why hospitals, practices, and medtech SMEs use patient-specific 3D implants: they plan complex procedures, reduce the risk of surprises in the OR, and can show patients very concretely what will happen. At the same time, several pitfalls lurk along the way – incorrect scan parameters, unclear responsibilities, overly optimistic assumptions about desktop printers. Further down, you will find our typical process from inquiry to the finished part, including specific settings and the mistakes we made ourselves at the beginning.

Source: Own Representation

Basics

When we talk about patient-specific 3D-printed implants, we mean components that fit the anatomy of an individual person exactly and are based on CT or MRI data. Typical examples include cranial plates for trauma or tumors, plates and drilling guides in orthopedics, dental implants with drilling guides, as well as orthoses and masks for positioning or radiation therapy. The general process is always the same: you start with image data from radiology, segment the relevant structures, create a 3D surface model from it (e.g., in STL format), and prepare it for 3D printing of an implant or a test model.

A patient-specific implant is always intended for one specific person – for example, a cranial plate that precisely closes the defect in the CT and must be neither larger nor smaller. In regulation, we also speak of custom-made devices (CMD): these are medical devices that are manufactured with patient-specific design characteristics based on a written prescription for exactly one patient and are not mass-produced. Patient-matched devices lie somewhere in between: they are produced in validated serial processes but are adapted to a patient's anatomy, for example, patient-specific plates from a large implant manufacturer. As soon as a part lands in the body or is used directly for diagnosis or therapy, we legally speak of a medical device. Then you automatically play by the rules of EU-MDR, national laws and relevant standards, for example in quality management. The Swiss regulatory authority Swissmedic summarizes this compactly in a fact sheet on 3D printers and medical devices – including references to relevant norms and standards ( ISO-Standards). Models printed purely for training or demonstration purposes are significantly simpler in terms of regulation, as long as it is clearly documented that they are not used for direct treatment decisions.

Preparation

Before you print the first layer, you should know quite precisely what should be on the table in the end. Is it a patient-specific implant, a surgical guide, an anatomical training model for students, or a demonstration object for patient consultations? The clearer the purpose, the easier the material selection, software decisions, and coordination with the hospital or customer.

In projects involving implants or surgical guides, you cannot avoid a clean setup with clear roles. Usually, you need a responsible physician, radiology for imaging, a medical technology team or manufacturer with an approved quality management system, and a documented assignment to EU-MDR oder nationalen Vorgaben. you can plan a bit more loosely for purely anatomical models, for example, in cooperation between the clinic, a university laboratory, and the maker scene – the only important thing is that it is documented in writing that the model is not a medical device.

Technically, you need image data in DICOM-Format, as a basis, mostly CT with slice thicknesses of 0.5 to 1 millimeter. Coarser scanned datasets quickly appear blocky in the 3D model and unnecessarily complicate work in design. For sensitive areas like the skull base or spine, a maximum slice thickness of 1 millimeter has proven effective in our projects. Many teams use open-source software like 3D Slicer or commercial solutions like Materialise Mimics; at 33d.ch, we regularly see both options in customer projects.

For 3D printing itself, depending on the goal, you decide between in-house production and a certified partner. For purely training and anatomical models, a cleanly calibrated FDM or resin printer is often sufficient. For implants or highly stressed guides, metals like titanium, high-performance polymers like PEEK or PEKK, and special medical resins are usually used – typically by a service provider precisely designed for these materials and standards.

Practical checklist from our workshop

In practice, the following sequence has proven effective for us before we even think about the slicer:

• Document the use case in writing together with the clinical team (purpose, region, desired product).
• Clarify responsibilities: Who is medically responsible, who is the manufacturer, who segments, who prints.
• Define the scan protocol (modality, slice thickness, field of view) and agree on it with radiology.
• Determine the software stack (e.g., DICOM viewer, segmentation, CAD, slicer) and check access.
• Decide early whether you are creating a medical device or a pure model and document this in the project brief.

If all of this is clearly defined, the actual design and printing steps will run much more smoothly – and you will have to improvise less later.

Step-by-step guide

The production of patient-specific 3D-printed implants and models almost always follows the same pattern for us. The details change depending on the specialty, but the logic remains the same.

Step 1: Define clinical use case and product type

Together with surgery and radiology, we first clarify what exactly the model or implant will be used for – for example, a patient-specific cranial plate after an accident, a drilling guide for dental implants, or a heart model for surgical planning. At the same time, we determine whether it is an implant, a surgical guide, or a pure anatomical model, as this affects its classification as a custom-made device, patient-matched device, or non-medical device. A good test: you can write the use case in one sentence, and all parties agree.

Step 2: Plan and perform imaging

For bony structures, we usually plan a CT scan, and for certain soft tissue applications, a high-resolution MRI. The parameters are important: slice thickness of max. 1 millimeter, appropriate reconstruction kernels, and a field of view that completely covers the relevant region. In practice, we repeatedly see datasets where half of the mandible is missing – this is annoying because you then have to rescan everything. Therefore, in the DICOM viewer, we consistently check whether the dataset is complete and artifact-free.

Step 3: Create segmentation and 3D surface model

The DICOM data then lands in segmentation software like 3D Slicer oder Mimics. Here, the target structures are marked – for example, the calvaria, alveolar ridge, or vertebral bodies – and exported as a 3D mesh, usually in STL format. We are well aware of typical pitfalls: holes in the mesh after strong metal artifact reduction, cut-off tips with too small a field of view, or stair artifacts with too large a slice thickness. We therefore always perform a brief visual check by overlaying the 3D model with the original images and comparing edges and contours.

Source: 3dprintingindustry.com

Schematic workflow for the production of a patient-specific 3D-printed bone implant with optimized lattice structure.

Step 4: Design implant or model

Based on the segmented anatomy, the actual design is created. For complex cases, we like to use medical design software like Materialise 3-matic Medical, , which allows for very fine control of lattice structures, screw holes, and transitions. For a cranial plate, for example, we define the contour along the defect boundaries, the plate thickness, and the position of the fixation points; for drilling guides, we position sleeves so that the angle and depth of the drill are precisely guided. Mistakes that happened to us: too thin bridges that break when removing supports, or geometries that are difficult to place in the OR. Therefore, today we work with clear minimum thicknesses and edge distances, which we further refine with each project.

Step 5: Choose printing strategy and material

For actual implants, we consistently rely on certified partners who produce titanium or PEEK implants in validated processes. For anatomical models and training objects, we print a lot ourselves – often with FDM or resin. Layer heights of 0.1 to 0.2 millimeters have proven effective for us, as the ratio of detail accuracy and printing time is good. A simple check is a reference measurement in the model, such as a 50-millimeter bar, which we measure after printing. If the deviation is less than one millimeter, that is more than sufficient for most planning purposes.

Material choice in quick comparison

Step 6: Postprocessing, quality assurance, and documentation

After printing, we remove support structures, clean the part, and prepare for sterilization depending on the intended use. For implants, this includes mechanical testing, dimensional checks, and formal release in the quality management system for the manufacturer; nothing works without a solid system behind the scenes. Manufacturers of custom-made devices must demonstrate a complete quality management according to EU-MDR . For training and anatomical models, a documented visual inspection, a comparison of actual vs. nominal values for selected dimensions, and brief feedback from users after use are often sufficient.

Step 7: Clinical application, feedback, and iteration

The most exciting moment is always the first use: does the plate fit as well as the virtual model promised? Is the drilling guide used intuitively, or does it get stuck in an unexpected place? After such cases, we specifically solicit feedback from the OR and document fit accuracy, handling, and any peculiarities. Over time, this leads to internal design rules and checklists that make subsequent projects significantly faster and safer. At 33d.ch, our current standard workflow for patient-specific projects has emerged precisely from this feedback process.

Common mistakes & solutions

Many difficulties are repeatedly encountered in projects – regardless of whether it's about cranial plates, dental guides, or orthoses. We see three typical stumbling blocks particularly often.

Mistake 1: Regulation comes into play too late. At the beginning, an implant project often seems like an exciting technical case, and suddenly the question arises: Who is actually the manufacturer in the sense of EU-MDR? If no one has a clear answer, that's a warning sign. Our solution: from the very start of the project, we define which partner acts as the manufacturer, how the release process works, and which documents must ultimately end up in the technical dossier.

Mistake 2: CT data is too coarse or incomplete. This also happened to us at the beginning: we had a beautiful dataset with 2-millimeter slices – until we rotated the model in the viewer and saw stair artifacts everywhere. Such data is hardly suitable as a basis for precise plates or guides. Today, we consistently require slice thicknesses of no more than 1 millimeter and briefly check each series in the viewer before anyone starts segmentation.

Mistake 3: Desktop printers are overestimated. For prototypes and training objects, we love our workshop printers, but they don't replace a qualified implant manufacturer with tested materials and validated processes. When someone says: "We can quickly print that in titanium in the basement," we know that a discussion about roles and responsibilities is needed first. Our rule of thumb: design and test models gladly in-house, everything that goes into the body belongs in a strictly regulated manufacturing process.

A nice example from practice: In an ENT project, 3D-printed sinus models were initially used purely as training objects. However, the material was so soft that the milling feel had little to do with reality – which frustrates experienced surgeons in particular. After a material change and adjusted wall thicknesses, the handling was significantly more realistic, and in a study, the models could even be meaningfully compared with cadavers. Such feedback loops are invaluable because they directly translate into better designs and material decisions.

Variations & adaptations

The process described above can be adapted to very different goals – as long as you know where you can be creative and where regulation sets clear boundaries. For patient-specific metal implants, many teams work with specialized manufacturers who produce a titanium or PEEK solution from the design and provide the regulatory documentation. For orthoses or positioning aids, for example in radiation therapy, you can also handle individual steps in-house, as long as the overall process is embedded in a tested quality management system.

3D-printed anatomical models as a supplement or replacement for cadavers in education are very exciting. Studies show that such models enable comparable or even better knowledge transfer in certain scenarios – and are simultaneously much easier to scale because you can reprint them as often as needed. For us, they are particularly useful when rare pathologies or complex variations are to be trained, which are difficult to find on a "standard cadaver."

For technology enthusiasts and makers who want to try their hand at anatomical models without immediately diving into the world of medical devices, it's worth taking a look at open platforms like the NIH 3D Print Exchange. There you will find tested anatomical models, molecular structures, and teaching models that are deliberately provided for education and research. The important thing is that it remains clear: these files are not automatically approved as implants or surgical guides – but they provide you with an excellent basis for learning, experimenting, and for your first own projects.

There is also a lot of activity in materials. In clinical practice, metals like titanium, biocompatible polymers like PEEK and PEKK, technical plastics, as well as silicones and resins currently dominate. In parallel, the community is researching hydrogel-based bioinks with living cells, which could potentially be used for tissue or organ structures. In our daily work, this is more of an exciting outlook – most projects still revolve around "classic" plastics and metals that can be reliably printed, cleaned, and documented.

If you want to see the workflow from CT data to the finished 3D model step-by-step, it's worth watching the embedded video. There you can see very well how DICOM import, segmentation, and model preparation interlock – exactly the steps we go through daily in our projects.

FAQ: Questions we repeatedly encounter in projects

Finally, we answer a few questions that we are regularly asked in our daily work at 33d.ch – whether on the phone, in the meeting room, or directly at the machine.

Question 1: Can I simply produce patient-specific implants with a good desktop 3D printer?

Short answer: no. Implants and surgical guides are subject to the requirements for medical devices – meaning quality assurance, material certifications, risk management, and often clinical evaluation. A desktop printer is great for prototypes, test parts, or training models, but it doesn't replace a certified manufacturing process with validated parameters and documented traceability. A sensible approach is: you develop the design and test it with your own printers, but have the actual implant produced and released by an authorized manufacturer.

Question 2: How fine do CT or MRI data need to be for 3D models to be meaningfully printable?

For bone, slice thicknesses of 0.5 to 1 millimeter have proven effective in our practice. Coarser slices create visible steps and cost you a lot of time in post-processing. Many teams get by well with 1 millimeter for surgical guides, while 1.25 millimeters are often already borderline. For very complex structures – such as the skull base or fine joint surfaces – a special 3D printing protocol in radiology, precisely tailored to your project, is worthwhile.

Question 3: What are the advantages of 3D-printed anatomical models in education compared to cadavers?

3D models are infinitely reproducible, do not require cooling, and can be specifically designed to highlight certain pathologies or variations. You can color-code, label, saw, drill – and then simply reprint the model. Especially for large cohorts in education or for recurring simulation training, such models are therefore very attractive. Our impression from projects with universities: students often dare to do more with printed models and repeat critical steps more frequently than with cadavers.

Question 4: What's the deal with bioprinting and biofilaments – is that already commonplace?

Bioprinting works with so-called bioinks, usually hydrogel-based carrier materials, in which living cells are embedded. These can be used in the laboratory to produce tissue structures, tumor models, or test systems for drugs. In clinical practice, we have hardly encountered this yet; titanium, PEEK, and various plastics still dominate there. If you want to start with patient-specific implants, it makes sense to focus on these established materials first and view bioprinting more as an exciting future topic.

Question 5: Where can I find reputable 3D models for training and patient education?

A very good point of contact is the NIH 3D Print Exchange. There you will find thousands of biomedical models – from organs to bones to molecules – as well as tools to create your own files. In parallel, many university libraries and medical technology laboratories maintain their own curated collections of 3D print resources specifically created for teaching and simulation. This allows you to work relatively quickly with high-quality datasets without having to build everything from scratch.

Mini-conclusion for your everyday life with patient-specific 3D implants

• Without a clearly defined use case and role distribution, every implant project becomes unnecessarily complicated – take your time for this at the beginning.
• Good image data is half the battle: clean DICOM scans with appropriate slice thickness save hours later in segmentation and design.
• Use your workshop printers for prototypes and training models, but rely on regulated manufacturing processes for actual implants.
• Feedback from surgery and education is not a "nice-to-have" but the engine for better designs, material decisions, and workflows.

If you keep this in mind, your first patient-specific project will not be an experiment, but the start of a repeatable process.

The second video shows how clinics and industry collaborate to implement patient-specific implants on a larger scale. If you want to see how your own workflow can be professionalized in the long term, this is a good source of inspiration.