Why detailed modelling matters in the pre-clinical evaluation of temporomandibular joint implants

This study demonstrates that while simplified finite element models of the mandible can preserve spatial stress-strain trends for preliminary implant design, they significantly underestimate peak stress and strain values compared to detailed, tissue-specific models, underscoring the necessity of high-fidelity modelling for final pre-clinical evaluation of temporomandibular joint implants.

The Gist

Imagine your jaw is a high-performance race car engine. When the engine breaks down completely (a condition called end-stage TMJ disorder), doctors have to replace the broken parts with a custom-made prosthetic joint. But before they can install this new part in a real person, they have to test it virtually to make sure it won't snap, crack, or cause pain.

This is where computer modeling comes in. Scientists build a digital twin of the jaw and run simulations to see how the new implant handles the stress of chewing, biting, and talking.

The big question this paper asks is: "How detailed does our digital twin need to be to get the right answer?"

The Three Levels of Detail

The researchers tested three different ways to build this digital jaw, using a "Goldilocks" approach:

1. The "Super-Detailed" Model (Model 1): This is like a hyper-realistic video game character. It includes every tiny part: the hard outer shell of the bone (cortical), the spongy inner bone (cancellous), the teeth, the little shock-absorbing pads around the teeth (periodontal ligaments), and the cartilage in the joint. It's accurate, but it's also computationally heavy—like trying to run a 4K movie on an old laptop. It takes a long time and a lot of power to simulate.
2. The "Simplified" Model (Model 2): This is like turning the video game graphics down to "Medium." The researchers made the entire jaw out of just the hard outer bone material. They ignored the spongy inside and the teeth. It's faster to run, but is it accurate?
3. The "Ultra-Simplified" Model (Model 3): This is the "Low Graphics" setting. They took Model 2 and removed the teeth entirely, leaving just a smooth bone structure. This is the fastest to run, but it's the furthest from reality.

The Experiment: The Chewing Test

The team put these three digital jaws through a rigorous workout. They simulated six different types of biting and clenching (like biting a sandwich, grinding your teeth, or chewing gum) and tested two types of implants: a "narrow" one and a "standard" one. They also tested two scenarios:

• Fresh Surgery: The implant is just sitting there, held by screws (like a new tire not yet broken in).
• Healed Surgery: The bone has grown into the implant, fusing them together (like a tire fully seated and broken in).

The Big Reveal

Here is what they found, using some simple analogies:

• The "Stiffness" Trap: When the researchers simplified the model (removing the spongy bone and teeth), the digital jaw became stiffer. Imagine replacing a springy mattress with a solid concrete slab. Because the concrete slab doesn't bend as much, it absorbs less stress.
• The Underestimation: Because the simplified models were "stiffer," they predicted that the bone and the implant would experience much less stress than they actually would.
  • For the bone, the simplified models underestimated the strain (stretching) by up to 50%. That's like saying a bridge can hold 100 tons when it can actually only hold 50.
  • For the implant itself, the stress was underestimated by up to 44%.
• The Good News (The Pattern): Even though the numbers were wrong, the patterns were right. The simplified models showed the stress in the same places as the detailed models. If the detailed model showed a weak spot near the screw, the simplified model showed a weak spot there too. It just thought the weak spot wasn't as weak as it really was.

The Verdict: When to Use Which?

The authors conclude that simplification is a tool, not a shortcut.

• For the "Sketching" Phase (Preliminary Design): If you are just brainstorming ideas and want to quickly test 50 different shapes of an implant, the simplified models are great. They are fast, cheap, and they tell you the general direction of the stress. They are like sketching a car design on a napkin.
• For the "Final Blueprint" Phase (Pre-Clinical Evaluation): Before you actually manufacture the implant and put it in a human, you must use the detailed model. Because the simplified models underestimate the stress, relying on them could lead to a design that looks safe on the computer but fails in the real world. You need the "4K video game" version to ensure the patient's safety.

In a Nutshell

Think of it like baking a cake.

• Simplified Models are like using a recipe that says "add some flour and sugar." It's fast, and you'll get something that looks like a cake, but it might be too dense or dry. It's fine for a quick experiment in the kitchen.
• Detailed Models are like weighing every gram of flour, sugar, and egg, and measuring the exact temperature of the oven. It takes longer and requires more effort, but it's the only way to guarantee the cake comes out perfect and safe to eat.

The takeaway: Don't skip the details when the stakes are high. If you are designing a life-changing medical implant, you need the full, detailed picture to avoid surprises later.

Technical Summary

1. Problem Statement

End-stage Temporomandibular Joint (TMJ) disorders often require total joint replacement (TJR) using patient-specific implants. The design and safety evaluation of these implants rely heavily on Finite Element (FE) analysis derived from patient CT scans.

• The Challenge: Generating highly detailed, subject-specific FE models (including cortical bone, cancellous bone, teeth, periodontal ligaments, and fibrocartilage) is computationally expensive and time-consuming.
• The Gap: To save time, many researchers simplify these models by treating the entire mandible as cortical bone or by excluding dental crowns. While these simplifications are known to alter mechanical behavior, there is no systematic quantitative investigation into how much these simplifications affect the pre-clinical evaluation of TMJ implants. This lack of data poses a risk of underestimating failure risks during the design phase.

2. Methodology

The study employed a comparative FE analysis approach to quantify the impact of geometric and material simplifications.

• Subject Data: Clinical CT scans of a healthy 20-year-old male were used to create a baseline model.
• Model Variations (Three Levels of Simplification):
  1. Model 1 (Detailed/Reference): A fully segmented mandible with tissue-specific material properties (cortical bone, cancellous bone, enamel, dentin, periodontal ligament, and articular fibrocartilage).
  2. Model 2 (Stage I Simplification): The entire mandible (including cancellous bone, teeth, and PDL) was homogenized as cortical bone only.
  3. Model 3 (Stage II Simplification): Derived from Model 2, but with dental crowns excluded entirely.
• Implant Configurations:
  • Two implant types: Narrow and Standard Biomet stock TMJ implants (45 mm).
  • Two interface conditions: Non-osseointegrated (NOI) (frictional contact, $\mu=0.3$) representing immediate post-op, and Osseointegrated (OI) (bonded contact) representing long-term healing.
  • Total Models: 15 FE models (3 intact + 12 implanted combinations).
• Loading Conditions:
  • Simulated a complete mastication cycle with six clenching tasks: Incisal (INC), Intercuspal (ICP), Right/Left Molar (RMOL/LMOL), and Right/Left Group (RGF/LGF) bites.
  • Muscle forces from seven major masticatory muscles were applied.
• Analysis Metrics:
  • Bone: Maximum principal strain (failure criterion for brittle bone).
  • Implants/Screws: Von Mises stress (failure criterion for ductile Ti-6Al-4V alloy).
  • Statistical Comparison: Linear regression and $R^2$ values were calculated between Model 1 and the simplified models (2 & 3) for the ICP loading case.

3. Key Contributions

• Quantification of Simplification Errors: The study provides the first systematic quantification of how mandibular simplifications (homogenizing bone and removing teeth) impact stress/strain predictions in TMJ implants.
• Differentiation of Design Stages: It establishes a clear boundary between when simplified models are acceptable (preliminary design) versus when detailed models are mandatory (final pre-clinical evaluation).
• Interface Analysis: It evaluates how these modeling errors persist or change under different biological conditions (immediate post-op vs. osseointegrated).

4. Key Results

• Strain Underestimation in Bone:
  • Simplified models significantly underestimated the maximum principal strain in the mandibular cortex.
  • Model 2 showed the lowest strains (range: 394–994 $\mu\epsilon$), while Model 1 (detailed) showed the highest (range: 610–1880 $\mu\epsilon$).
  • The reduction in predicted strain was up to 50% in simplified models compared to the detailed model.
  • Reason: Homogenizing the mandible as cortical bone increases overall stiffness, reducing predicted deformation.
• Stress Underestimation in Implants:
  • Simplified models also underestimated von Mises stresses in the implants and screws.
  • Reductions ranged from 2.4% to 44.1% for implants and 0.1% to 48.8% for screws.
  • Model 1 consistently predicted the highest stresses, followed by Model 2, then Model 3.
• Spatial Trends vs. Magnitude:
  • While the magnitude of stress/strain was significantly lower in simplified models, the spatial distribution patterns (where the high stress occurs) remained largely consistent.
  • Regression analysis showed strong correlation ($R^2$ up to 0.97) for implant stresses, but lower correlation for bone strains due to outliers in complex geometries (e.g., coronoid process, screw interfaces).
• Osseointegration Effect:
  • All models correctly predicted that osseointegration reduces stress and strain compared to the non-osseointegrated state, though the magnitude of this reduction varied by model complexity.

5. Significance and Conclusion

• Preliminary Design: Simplified models (Model 2 and 3) are computationally efficient and preserve the qualitative trends of stress distribution. They are suitable for preliminary parametric design and screening of implant geometries.
• Final Evaluation: The significant underestimation of stress magnitudes (up to 50%) in simplified models means they cannot be relied upon for final safety assessments. A detailed FE model (Model 1) is essential for the final pre-clinical evaluation of TMJ implants before fabrication and clinical trials to ensure patient safety and prevent implant failure.
• Clinical Implication: Relying on simplified models for final validation could lead to the approval of implants that are actually at risk of fatigue failure or bone resorption under physiological loads.

In summary, the paper argues that while simplification is a useful tool for early-stage design optimization, the complexity of the mandible's heterogeneous structure is critical for accurate biomechanical prediction in the final stages of TMJ implant development.