Braiding is one of the oldest ways to make something strong and flexible at once, and it is about to get a new manufacturing route. Researchers led by IMDEA Materials Institute and the Universidad Politécnica de Madrid (UPM), working with MIT and the University of Galway, have described a way to 3D print braided, superelastic nickel-titanium structures for vascular implants. Their paper, "Additive manufacturing of superelastic NiTi braided metamaterials," published in npj Metamaterials, sets out a patented single-step process. It produces woven-style nitinol devices that can be shaped to a patient's anatomy, and they fold and spring back without the brittleness of conventional printed lattices.

The paper itself appeared a few weeks ago, on September 30. IMDEA issued its release on October 13, and the work reached the wider additive manufacturing audience this week, when trade outlets covered it on October 19 and 20. It is lab work, not a product: the team says the devices have not been tested in living tissue yet.

The Problem With Printed Lattices

Nitinol is the standard material for self-expanding stents and other transcatheter devices because it is superelastic. It can be squeezed into a catheter and then recover its shape once deployed. Printing nitinol is attractive for one obvious reason: every patient's vessels are different, and a printed device could in principle be made to fit them.

The difficulty is geometry. Most printed stent designs so far are lattices, rigid networks of struts joined at nodes. According to VoxelMatters' coverage, the researchers found that compressed lattices become excessively stiff, exert high forces on arterial walls and risk fracture at strains approaching 25%. That is a serious constraint for a device that must be crimped into a delivery catheter and then sit inside a pulsing artery for years.

Braids behave differently. Instead of bending rigid struts, they accommodate deformation through the movement and reorientation of interwoven filaments, much as conventional braided stents do. The IMDEA/UPM team set out to print that structure directly in nitinol, rather than weaving it from individual wires.

How the Process Works

The npj Metamaterials abstract describes an algorithmic design pipeline. It generates both lattice and woven structures and adjusts them to anatomical surfaces. The software can also build stiffness gradients into a single part and make transitions from lattice regions to woven regions within the same device. That matters for implants that need a firm anchoring zone in one place and compliance in another.

The printed parts are not used as-built. IMDEA's release says the team post-processed them with chemical etching followed by electropolishing to reach the surface finish required for implantation. That cut fibre thickness by 47.7% and cross-sectional area by 71.1% compared with the as-printed state. At implant scale the surface finish is not cosmetic, and the abstract says these post-processing steps also contribute to achieving the superelasticity the devices need.

3D Printing Industry reports that the method is patented as a one-step process, and that the work has received a National James Dyson Award. It also builds on an earlier IMDEA/UPM study of woven nitinol, published in February 2026 in Virtual and Physical Prototyping.

The Numbers

The headline figures come from IMDEA's release:

  • Compression: the braided stents withstood compression to 50% of their original diameter.
  • Recovery: they showed more than 90% elastic recovery after that compression.
  • Chronic outward force: up to five times lower than conventional lattices.

The last figure needs some context. Chronic outward force is the sustained pressure a self-expanding device keeps applying to the vessel wall once it is in place. Lowering it, the institute says, reduces the risk of vascular injury and thrombosis. Cutting it by as much as a factor of five, while still surviving 50% compression, is the main argument for braids over lattices here. Note that the two deformation figures are not a like-for-like comparison: the 25% is the strain at which lattices risk fracture, while the 50% is compression of the stent's diameter.

The validation was also narrow. The team checked several prototypes against standard mechanical performance indicators using a single compression-expansion cycle that replicates loading into a catheter and release. That is one cycle, not years of pulsing in an artery.

What the Devices Could Become

The paper's abstract lists several proof-of-concept prototypes. They include stents, transcatheter aortic valve replacements, customized clot retrievers for thrombectomy, tapered flow diverters, braided bifurcations and fenestrated stents. In the body of the paper, the valve frame is the clearest example of the personalisation argument: start from a patient's aortic root as derived from CT scans, generate a fitted braided geometry algorithmically, and print it.

Neurovascular devices are a further target, but VoxelMatters notes they will need more work. Reaching remote, hard-to-access regions of the brain would require further miniaturization of the nitinol structures, with technologies such as two-photon polymerization using metallic suspensions, plus magnetic functionalization for remote magnetic navigation. That suggests a different fabrication route from the current metal process for the smallest devices.

What It Does Not Show Yet

All of this is bench-top. 3D Printing Industry reports that the structures have not been tested in living tissue. In vitro biocompatibility testing and then animal studies must come before any patient trials. The mechanical numbers are promising, but implants fail for biological reasons as well as mechanical ones, and none of those questions have been answered yet. The performance figures in this article come from the institute's release and trade coverage, which match one another.

What It Means for Makers

No one is printing nitinol stents on a desktop machine, and that will not change. But the design idea here applies well beyond medicine, and it is worth taking in.

The main lesson is topology, not material. A lattice and a braid can look alike at a glance, yet they behave very differently under large strain. Lattices flex by bending rigid struts. Interwoven fibres that can move relative to one another take up the deformation by reorienting, and the structure recovers. Makers who design compliant mechanisms, flexible printed fabrics or energy-absorbing structures in TPU or nylon are dealing with a similar trade-off, though this study tested nitinol only. This research makes the case that "woven" and "lattice" are separate design categories, not variations on one.

The algorithmic side is just as relevant. Fitting a structure to a scanned surface, with stiffness gradients and changes between structure types in one part, is the direction generative design tools have been heading. The IMDEA/UPM work is a high-stakes example of why that matters.

Post-processing also turns out to be part of the design. Etching away nearly half of each fibre's thickness after printing is not cleanup. It is part of how the part reaches its final properties. Anyone tuning flexures in printed parts will recognise the point: what comes off the build plate is often not the finished part.

Sources