Virtual Population to Re-assess AAA Risk Using Neck Geometry and Shape Compactness Alongside Maximum Diameter

This study introduces an automated framework for generating demographically stratified virtual abdominal aortic aneurysm populations to demonstrate that proximal neck diameter and shape compactness, rather than maximum diameter alone, are the primary determinants of aneurysm haemodynamics.

The Gist

Imagine your body's main highway, the aorta, is a giant garden hose. Sometimes, a weak spot in that hose swells up into a balloon-like bulge. This is called an Abdominal Aortic Aneurysm (AAA). If that balloon pops, it's a medical emergency.

Right now, doctors have a simple rule for deciding when to fix these balloons: measure the widest part. If the balloon gets bigger than a specific size (like 5.5 cm for men), they operate. But here's the problem: sometimes balloons smaller than that pop, and sometimes huge ones don't. The "size rule" isn't perfect.

This paper introduces a new way to understand these balloons using computer magic instead of just a ruler.

The Problem: We Need More Data

To figure out why some balloons are dangerous and others aren't, scientists usually need to study thousands of real patients. But real patient data is hard to get (privacy issues, not enough women in studies, etc.). It's like trying to predict the weather by only looking at one day's forecast.

The Solution: The "Virtual Population" Factory

The researchers built a digital factory that can print out thousands of fake, but scientifically accurate, AAA models. Think of it like a video game character creator, but instead of making cool avatars, it makes realistic blood vessels.

1. The Blueprint: They started with real scans from 258 patients.
2. The Mixer: They used math to mix and match different features (like neck width, balloon size, and how curvy the hose is) to create 182 new, unique virtual patients.
3. The Stress Test: They ran 364 computer simulations (like wind tunnel tests for blood flow) on these virtual patients to see how the blood pushes against the walls.

The Big Discoveries: It's Not Just About Size

When they looked at the results, they found that the "widest part" rule was missing the real story. Here are the three main things they found, explained with analogies:

1. The Neck is the Real Boss

Imagine the aneurysm is a cave. The "neck" is the tunnel leading into it.

• The Finding: The width of the tunnel (neck) matters way more than the size of the cave itself.
• The Analogy: If you have a wide tunnel leading into a cave, the wind (blood) shoots in like a firehose, slamming hard against the walls. This creates high stress. If the tunnel is narrow, the wind slows down and swirls gently.
• Takeaway: Doctors should look closely at the neck, not just the biggest part of the bulge.

2. Shape Matters: The "Roundness" Factor

• The Finding: How "round" or "bumpy" the balloon is changes how the blood moves.
• The Analogy: Think of a smooth, round beach ball versus a crumpled potato chip.
  • A smooth, round shape (high "sphericity") lets the blood flow smoothly, like water in a calm river. It reduces dangerous swirling.
  • A bumpy, weird shape (low "convexity") creates chaotic eddies and whirlpools, which can wear down the vessel wall.
• Takeaway: A perfectly round, large aneurysm might actually be safer than a smaller, lumpy, weird-shaped one.

3. Size is a "Low-Speed" Zone

• The Finding: Making the balloon bigger doesn't necessarily make the peak pressure higher, but it does create a bigger area where the blood moves very slowly.
• The Analogy: Imagine a wide, slow-moving river. The water isn't crashing hard against the banks (low peak stress), but because it's moving so slowly, mud and trash (blood clots) start to pile up on the bottom.
• Takeaway: Big aneurysms are dangerous not because they explode from pressure, but because they become breeding grounds for clots and decay due to slow blood flow.

Why This Matters

This study is like upgrading from a ruler to a 3D scanner.

• Old Way: "Is the balloon bigger than 5.5cm? Yes? Cut it out."
• New Way: "Let's look at the neck width, the shape, and how the blood swirls. Even if it's small, if the neck is wide and the shape is weird, it might be risky. If it's big but round and calm, maybe we can wait."

The Bottom Line

The researchers created a "Virtual Population" to prove that geometry is destiny. By looking at the shape and the neck of the aneurysm, not just its size, doctors might be able to save more lives by catching the dangerous ones earlier and avoiding unnecessary surgery on the safe ones. It's a step toward truly personalized medicine, where every patient gets a risk assessment tailored to their unique anatomy.

Technical Summary

1. Problem Statement

Abdominal Aortic Aneurysm (AAA) rupture risk is currently assessed primarily based on the maximum transverse diameter ($D_{max}$), with surgical intervention typically recommended when $D_{max}$ exceeds 55 mm (men) or 50 mm (women). However, clinical evidence shows significant limitations to this size-based metric:

• False Negatives: A significant percentage of ruptures occur below the surgical threshold, particularly in women.
• False Positives: Many large aneurysms do not rupture and may not require immediate intervention.
• Data Scarcity: Existing Computational Fluid Dynamics (CFD) studies rely on small, often male-dominated cohorts (e.g., <100 patients), lacking statistical power for subgroup analysis (age/gender) and failing to capture the full spectrum of anatomical variability.
• Methodological Gaps: Current synthetic data generation methods (GANs, VAEs) often lack anatomical interpretability or user control, while interactive morphing tools are too labor-intensive for population-scale studies.

The study aims to overcome these limitations by creating a scalable, automated framework to generate a demographically stratified virtual population of AAAs to systematically investigate the relationship between complex geometric features and hemodynamic biomarkers.

2. Methodology

The authors developed a fully automated, constraint-aware framework to generate synthetic AAA geometries and perform high-throughput CFD simulations.

A. Data Foundation and Stratification

• Source Data: 258 pre-operative CTA scans from the Kyriakou et al. dataset (222 males, 34 females).
• Stratification: The dataset was stratified by gender and four age brackets (50–59, 60–69, 70–79, 80+).
• Key Parameters: Four primary anatomical diameters were defined: Neck Diameter 1 ($D_{neck1}$), Neck Diameter 2 ($D_{neck2}$), Maximum Diameter ($D_{max}$), and Distal Diameter ($D_{distal}$).

B. Geometric Generation Pipeline

1. Statistical Sampling: Probability distributions (Log-normal, Gamma, Weibull) were fitted to the anatomical parameters for each demographic subgroup. Random variates were sampled within clinically plausible ranges.
2. Constraint Enforcement:
   • Anatomical Rules: Ensured $D_{max} > D_{neck}$ and all diameters were positive.
   • Population Bounds: A Convex Hull analysis was performed on pairwise parameter combinations (e.g., $D_{neck1}$ vs. $D_{neck2}$). Generated geometries were validated to ensure they fell within the "universal interior set" of observed patient data, filtering out anatomically unrealistic combinations.
3. Morphing and Perturbation:
   • Stochastic Roughness: Latin Hypercube Sampling (LHS) introduced small-scale wall roughness.
   • Shape Morphing: A spherical control-point-based deformation algorithm introduced large-scale asymmetries and regional deformations while preserving inlet/outlet boundary conditions.
   • Validation: Generated variants were validated against the original patient distributions using Kullback-Leibler (KL) divergence (threshold < 0.3).

C. Computational Fluid Dynamics (CFD)

• Solver: OpenFOAM v9 using the pisoFOAM transient solver.
• Physics: Incompressible Newtonian flow with pulsatile, laminar inlet conditions (Fraser waveform).
• Boundary Conditions: Rigid walls with no-slip; zero pressure at the outlet. Two inlet profiles were tested: Plug (uniform) and Parabolic (fully developed).
• Mesh: Average mesh size of ~1.3 million cells with $y^+ \lesssim 1$.
• Output: 364 simulations were run on 182 validated geometries (two inlet profiles per geometry).

D. Metrics Extraction

• Geometric Descriptors (11): Including $D_{neck1,2}$, $D_{max}$, $D_{distal}$, Volume, Surface Area, Tortuosity, Sphericity ($\Psi$), Convexity ($\kappa$), and Average Radius.
• Hemodynamic Biomarkers (6): Mean Wall Shear Stress ($WSS_{mean}$), 95th percentile peak WSS ($WSS_{0.95}$), Time-Averaged WSS (TAWSS) low-shear area, Oscillatory Shear Index (OSI) mean, and OSI high-area percentage.

3. Key Contributions

1. Automated Virtual Population Framework: The first fully automated, constraint-aware pipeline to generate statistically significant, demographically stratified AAA cohorts for in-silico trials, bridging the gap between small patient studies and population-scale analysis.
2. Integration of Advanced Shape Metrics: Moving beyond simple diameters, the study incorporates Sphericity and Convexity (shape compactness) as critical descriptors, which have been under-explored in AAA literature but are standard in cerebral aneurysm research.
3. Systematic Correlation Analysis: A comprehensive correlation study linking specific geometric features to hemodynamic outcomes across a large cohort (182 geometries), revealing non-intuitive relationships that challenge current clinical paradigms.

4. Key Results

A. Dominant Determinants of Hemodynamics

• Neck Diameter 1 ($D_{neck1}$): The strongest predictor of wall shear stress.
  • Positive correlation with Mean WSS ($r \approx 0.77$) and Peak WSS ($r \approx 0.58$).
  • Negative correlation with low-TAWSS area ($r \approx -0.36$).
  • Mechanism: A wider neck allows a coherent, high-momentum jet to impinge directly on the sac wall.
• Maximum Diameter ($D_{max}$): Surprisingly, $D_{max}$ has minimal effect on peak shear ($r \approx -0.03$).
  • It primarily increases the area of low shear ($r \approx +0.20$).
  • Mechanism: A larger sac acts as a diffuser, slowing the flow and expanding the recirculation zone, thereby increasing thrombosis risk without increasing peak stress.
• Shape Compactness (Sphericity & Convexity): These are the strongest predictors of flow oscillation.
  • Strong inverse correlation with OSI ($r \approx -0.68$ for Sphericity, $-0.65$ for Convexity).
  • Mechanism: Compact (spherical/convex) sacs trap vortices centrally, reducing wall oscillation. Elongated or irregular shapes promote wall impingement and oscillatory flow.

B. Demographic and Sensitivity Findings

• Inlet Profile Sensitivity: While plug-flow profiles weakened some correlations, the dominant trends (Neck 1 driving WSS; Compactness driving OSI) remained consistent across both inlet types.
• Demographic Trends: The synthetic cohort successfully reproduced known trends: $D_{max}$ increases with age, and female aneurysms are consistently smaller in all dimensions than male aneurysms.

5. Significance and Clinical Implications

• Rethinking Risk Stratification: The study suggests that relying solely on $D_{max}$ is insufficient. Risk models should integrate proximal neck geometry (to assess peak shear) and shape compactness (to assess oscillatory shear and thrombosis risk).
• Personalized Medicine: The framework enables the creation of patient-specific risk profiles by mapping individual geometry to hemodynamic outcomes, potentially identifying high-risk patients with smaller aneurysms or low-risk patients with larger ones.
• Future AI/ML Applications: The generated dataset of 182 validated geometries with paired CFD data serves as a robust training set for developing machine learning surrogates that can predict hemodynamics instantly from 3D imaging, bypassing the need for time-consuming CFD simulations.
• Device Design: The findings highlight the need for endografts that account for neck variability and shape compactness, particularly for female patients who are currently under-represented in device sizing data.

In conclusion, this research provides a scalable, data-driven methodology to decouple the effects of size, shape, and neck geometry on AAA hemodynamics, offering a more nuanced basis for clinical decision-making than current diameter-based guidelines.