Sources of Variability in Intracranial Aneurysm Model Reconstruction and Evaluation: A Systematic Investigation

Area of Science:

• Medical imaging analysis
• Computational fluid dynamics
• Neurosurgery

Background:

• Accurate intracranial aneurysm (IA) model reconstruction is vital for understanding disease progression and performing computational simulations.
• Variability in reconstruction can affect diagnostic accuracy and treatment planning.

Purpose of the Study:

• To quantify the impact of segmentation thresholds and software platforms on IA geometry reconstruction.
• To assess the influence of inter-user variability on IA morphology evaluation.

Main Methods:

• Reconstructed 600 IA models from 100 DSA datasets using Materialise Mimics and 3D Slicer.
• Applied three grey value (GV) thresholds (1000, 1500, 2500) for segmentation.
• Performed geometric measurements by three users, comparing results against baseline parameters.

Main Results:

• Significant geometric variations observed across software platforms and thresholds (p < .05).
• 3D Slicer models were consistently smaller than Mimics models (-1.27% to -4.38%).
• Lower GV thresholds resulted in larger models (up to 15.9% increase in diameter).
• User variability was most pronounced in less experienced users (up to 22.67% deviation).

Conclusions:

• Segmentation software, threshold choice, and user interaction introduce significant variability in IA model geometry.
• Standardization of segmentation protocols, including threshold values and operator training, is essential for improving reproducibility in IA analysis.