Development of a new geometric severity score for nasal airway obstruction: A simulation-free approach to predict airflow resistance from CT geometry.

Abstract

Nasal airway obstruction (NAO) from septal deviation and inferior turbinate hypertrophy significantly impairs respiratory function, yet objective pre-operative assessment currently relies on Computational Fluid Dynamics (CFD), a technique too resource-intensive for routine clinical use. We present a geometric severity score (S) derived exclusively from CT image processing that predicts unilateral airflow resistance without requiring flow simulation. Six cross-sectional shape descriptors: area, circularity, perimeter-to-area ratio, DFT-based irregularity, solidity, and fractal dimension, are extracted from coronal slices, z-score normalised against a healthy reference group, and combined via an optimised weighted sum. A multiplicative exponential model with power-law flow-rate scaling, fitted jointly with metric weights using stochastic global optimisation, achieved R²=0.750 against CFD-derived unilateral resistance across two flow rates (15 and 30 L/min) in a cohort of two healthy controls and four patients with septal deviation. The score is anchored such that S=0 corresponds to healthy anatomy, S>0 to obstruction, and S<0 to over-patency. Cross-sectional area (r=-0.765) and perimeter-to-area ratio (r=0.779) showed the strongest individual correlations with resistance, while the composite score outperformed any single metric. This approach provides clinicians with a simulation-free tool for objective severity stratification from routine CT imaging.