Case Analysis: 3D Printing for Endoluminal Isolation of an Abdominal Aortic Aneurysm

The patient is a 68-year-old woman with an abdominal aortic aneurysm, intratumoral thrombus, bilateral common iliac artery aneurysm, and early stenosis of both renal arteries. For a covered stent-graft procedure, the surgeon needs to know two things before entering the room: where the proximal landing zone sits relative to the renal ostia, and how the iliac limbs will track through a calcified, tortuous access path. A patient-specific model answers both questions with a part you can hold and measure.

The data path starts with contrast CT. Slice thickness of 0.6–1.0 mm is the practical floor; thinner slices help at the renal ostia but multiply segmentation time. Segmentation means separating contrast-filled lumen from thrombus and from vessel wall. This is the step where most error enters. Thrombus is not contrast-enhanced, so a threshold-based mask will drop it and leave a narrowed lumen. We segment the outer wall and the lumen as two separate masks, then subtract to keep the thrombus volume visible in the print.

After segmentation comes wall thickness. A 2 mm nominal wall holds shape under handling. Below 1.2 mm the model flexes and your landing-zone measurements drift. Above 3 mm the renal ostia get bridged and the model stops representing the access problem you are trying to solve. We print the aortic body and both iliac limbs as one piece with a split line at the distal iliac, so the surgeon can test both a stiff and a flexible limb route.

• 1
  Slice thickness0.6–1.0 mm CT slices for aortic work; 0.6 mm at the renal ostia
• 2
  Two masksLumen and outer wall segmented separately, thrombus kept
• 3
  Wall thickness2 mm nominal; keep between 1.2 mm and 3 mm

Not every model needs the same resin. The decision turns on what the model is for. If the goal is sizing the stent-graft and checking the proximal neck length, a rigid material is correct because it will not compress under calipers. If the goal is rehearsing catheter tracking through a tortuous iliac, you want a compliant material that bends like the vessel. One model rarely does both jobs well, so we usually quote a rigid aortic body and a flexible iliac insert.

For rigid work we use opaque photopolymer with a Shore D hardness above 80. It holds a Ø0.5 mm calcification feature and accepts a 0.1 mm caliper without denting. For flexible limbs, Shore A 40–70 urethane-type resin tracks a guidewire without tearing at the bifurcation. Transparent resin is a third option and is used when the surgeon wants to watch a balloon inflate inside the model, but it scratches and clouds fast, so it is a one-use item.

Sterilization matters here too. Printed vascular models are not implanted, but they do enter the sterile field. Autoclave at 134 °C will deform most photopolymers. Ethylene oxide or hydrogen peroxide plasma is the safer route, and both need a material that will not craze. If a hospital requires autoclave, we switch to a high-temperature resin or machine the model from PEEK or POM instead.

• 1
  RigidShore D > 80 photopolymer for sizing and caliper work
• 2
  FlexibleShore A 40–70 resin for catheter and guidewire rehearsal
• 3
  TransparentVisualization only; clouds quickly and is single-use
• 4
  SterilizationEtO or H2O2 plasma; autoclave needs PEEK or POM

A printed model is only as good as three things: the scan, the segmentation, and the machine. Scan resolution sets the ceiling. Segmentation sets the offset. The printer, at 0.05–0.1 mm layer height, is usually the smallest error source. Engineers often blame the printer when the real gap sits upstream in the mask.

We check accuracy against the source CT, not against a CAD file. Fiducial pins are printed into the base plate and located on the model with a coordinate measuring machine. Landmark checks include the proximal neck diameter, the distance from the lowest renal artery to the aneurysm sac, and the iliac bifurcation angle. On a well-segmented case, these landmarks land within ±0.2 mm of the CT measurement. That number, not the machine tolerance, is the one to put in the model report.

This is also where the case analysis 3d printing workflow differs from a general prototype job. A prototype is judged on fit and appearance. A vascular model is judged on whether a measurement taken from it will hold inside the patient. Every tolerance in the build plan is chosen to protect that one property.

When the model is used as a bench fixture for stent-graft trimming, the trimming template itself is often machined rather than printed. A 6061-T6 or 316L template holds a sharp edge and a repeatable stop face, which a printed part will not. That is the point where the job moves from additive to subtractive.

• 1
  Scan0.6–1.0 mm slices set the accuracy ceiling
• 2
  SegmentationLargest single error source; thrombus must be kept
• 3
  VerificationFiducial pins checked on a CMM against source CT
• 4
  TemplateTrimming guides machined in 6061-T6 or 316L

A model that has not been measured is a prop. Our inspection sequence for a vascular model runs in three passes. First, raw material check: confirm the resin lot and cure profile, since an under-cured part creeps under load and will not hold a dimension overnight. Second, in-process monitoring: layer height, build orientation, and support placement are logged because support scars on the intimal surface will show up as false features. Third, final inspection: fiducial pins and anatomical landmarks are measured and reported.

Orientation deserves a note. Printing the aorta upright saves support on the lumen but puts layer lines perpendicular to the proximal neck, which is exactly where the surgeon measures. Laying the model at an angle so the neck is parallel to the build plate puts layer lines along the measurement direction instead. It costs more support and more cleanup time. We take that cost on sizing-critical cases.

Reports are issued on request. A typical model report lists the source scan, segmentation method, material and lot, build orientation, landmark measurements with uncertainty, and the sterilization method used. If the hospital's quality system needs a device master record, that report is what feeds it.

One limitation worth stating plainly. A printed model reproduces geometry, not tissue mechanics. It will not tell you how much a calcified neck will dilate under balloon pressure, and it will not predict endoleak. Treat it as a geometric reference and a rehearsal tool, not a physiological simulation.

• 1
  Raw materialResin lot and cure profile logged; under-cured parts creep
• 2
  OrientationAngle the model so layer lines run along the neck axis
• 3
  Final inspectionFiducial pins and landmarks measured, report on request

The printed model is one part of a larger instrument set. Hospitals and device teams often need a rigid sizing gauge, a trimming template, or a bench fixture that a printed part cannot provide. These are machined. On a 5-axis center we hold ±0.005 mm on gauge features, which is far tighter than any clinical measurement needs, but it means the gauge itself adds no error to the chain.

Materials for these fixtures follow the sterilization method. 316L stainless handles repeated autoclave cycles and stays dimensionally stable. 6061-T6 aluminum is lighter and cheaper but needs anodizing if it will be wiped with aggressive disinfectants. PEEK and POM are used when the fixture must not scratch a stent-graft or when radiolucency matters.

Surface finish also has a job here. A gauge that slides along a graft should sit at Ra 0.2–0.8 μm so it does not abrade the fabric. A handling fixture only needs Ra 1.6–3.2 μm as machined. We finish to the function, not to a blanket spec.

For teams that need both a printed anatomical model and machined fixtures from the same dataset, keeping them in one shop avoids a second round of data translation and a second confidentiality review. We work under NDA on request, and uploads stay confidential.

• 1
  316LRepeated autoclave cycles, dimensionally stable
• 2
  6061-T6Light and cheap; anodize if aggressive disinfectants are used
• 3
  PEEK and POMNon-marring and radiolucent fixture options
• 4
  FinishRa 0.2–0.8 μm on graft-contact faces