Geometric morphometrics approach for predicting olfactory region accessibility

Priya Vishnumurthy^{1, 2}, Thomas Radulesco³, Gilles Bouchet², Justin Michel³ & Alain Regard¹

¹ Nemera La Verpillière, Lyon, France

² Aix Marseille Univ, CNRS, IUSTI, Marseille, France

³Aix Marseille University, APHM, CNRS, IUSTI, La Conception University Hospital, Oto-Rhino-Laryngology and Head and Neck Surgery Department, Marseille, France

Summary

The variability of the nasal cavity significantly impacts the efficiency of olfactory drug delivery. While several studies have explored anatomical variability using average models or 2D measures, to our knowledge, none have addressed 3D shape variation in the region that drug particles must cross to reach the olfactory zone, namely the region of interest (ROI).

A geometric morphometric approach was used to assess shape variability of the ROI of 78 left and 73 right unilateral nasal cavities segmented from craniofacial CT scans. Using Viewbox 4.0, 10 fixed landmarks were digitized on each cavity. Thin Plate Spline (TPS) transformation with bending energy minimization was used to project 200 sliding semi-landmarks from a template to the 151 target cavities. Landmarks coordinates were standardised using Generalized Procrustes Analysis (GPA). Principal Component Analysis (PCA) was performed to identify dominant axes of shape variation. Morphological clusters were identified via Hierarchical Clustering on Principal Components (HCPC), conducted on the first five principal components. Inter-cluster differences were statistically tested (MANOVA, ANOVA, Tukey).

Three morphological clusters were identified. Cluster 1 exhibited a broader and shallower configuration, likely favouring olfactory accessibility. Cluster 3 was narrower and deeper, potentially less favourable for olfactory deposition. Cluster 2 showed intermediate traits. 31.5% of the 73 patients having both left and right cavities have at least one cavity pertaining to cluster 1.

Our results revealed distinct morphotypes of the ROI, which may influence olfactory region accessibility. These findings provide foundations for future computational deposition studies to evaluate how shape affects therapeutic targeting efficiency.