{
  "abstract": "Introduction/Purpose Computational fluid dynamics (CFD) is widely used to study intracranial aneurysm hemodynamics. Clinical translation remains limited due to concerns regarding reproducibility. One modeling assumption that is rarely justified is the number of cardiac cycles required for transient simulations to reach periodic convergence. It is unclear whether commonly used simulation durations are sufficient across different aneurysm geometries.Materials and Methods Three patient-specific intracranial aneurysm geometries located in the anterior cerebral artery (ACA), middle cerebral artery (MCA), and internal carotid artery (ICA) were simulated under standardized CFD conditions. Transient simulations were performed using zero initial conditions for ten cardiac cycles and steady-state initialization for three cardiac cycles. Periodic convergence was evaluated using surface-averaged wall shear stress (WSS). Metrics included cycle-to-cycle variation, waveform agreement, and root-mean-square error relative to the final cardiac cycle. Secondary hemodynamic metrics included oscillatory shear index, relative residence time, and velocity.Results Convergence behavior differed markedly by aneurysm geometry. The ACA aneurysm achieved stable surface-averaged WSS immediately after the first cardiac cycle, with near identical values (5.88 Pa) across all subsequent cycles. The MCA aneurysm demonstrated small but persistent cycle-to-cycle oscillations in surface-averaged WSS (6.05-6.06 Pa). In contrast, the ICA aneurysm exhibited sustained variability in surface-averaged WSS (7.15-7.38 Pa), with waveform root-mean-square error exceeding 5% in late cycles and transient peak errors exceeding 20%. Steady-state initialization did not improve convergence. Relative to the ACA and MCA aneurysms, the ICA aneurysm exhibited higher mean velocity (0.227 m/s vs. 0.179-0.189 m/s), higher surface-averaged oscillatory shear index (0.047 vs. 0.015-0.028), and greater dispersion in relative residence time (RRT), reflected by a 90th-percentile-to-median RRT ratio of 8.18 compared with 2.5-2.7.Conclusion Periodic convergence in intracranial aneurysm CFD is strongly dependent on aneurysm geometry and may not be achieved uniformly using commonly applied simulation durations. Standard practices using two to three cardiac cycles may be sufficient for some geometries but inadequate for others. Geometry-dependent convergence represents an underrecognized source of variability that may contribute to inconsistent results and limited reproducibility. Explicit assessment of convergence may be necessary to improve reliability and clinical translation of CFD studies.Disclosures B. Carlson: None. D. Patra: None.Abstract E-281 Figure 1Geometry-dependent convergence of surface-averaged wall shear stress. (A) Cycle-averaged surface-averaged wall shear stress across cardiac cycles for each aneurysm geometry. (B) Root-mean-square error relative to the final cardiac cycle, demonstrating persistent nonconvergence in the internal carotid artery aneurysm",
  "authors": [
    {
      "affiliations": [
        "Mayo Clinic Alix School of Medicine, Mayo Clinic, Phoenix, AZ"
      ],
      "name": "B Carlson"
    },
    {
      "affiliations": [
        "Neurosurgery, Mayo Clinic Arizona, Phoenix, AZ"
      ],
      "name": "A Bathini"
    },
    {
      "affiliations": [
        "Department of Neurological Surgery, Indiana University, Indianapolis, IN"
      ],
      "name": "D Patra"
    }
  ],
  "title": "E-281 Geometry-dependent periodic convergence in intracranial aneurysm computational fluid dynamics and implications for reproducibility",
  "uid": "ddd8fbba-45ff-5598-85b3-ea3b39cc10a9"
}