Measuring the Perceived Morphological Complexity of Intracranial Aneurysms
The rupture of intracranial aneurysms is a potentially deadly event. Aneurysm detection is based on imaging techniques, amongst which 3D angiograms deliver the highest quality. Currently, neuroradiologists assess such images only in a qualitative manner, even though it remains unclear how morphological properties relate to the disease status quantitatively.
In an effort to identify suitable descriptors that capture the irregularity of aneurysms, we relate a range of shape descriptors to expert assessments of aneurysm morphology. In a first step, we extracted 3D models of aneurysms from 3D angiographies and calculated geometry indices and moment invariants describing size and shape of aneurysms and surrounding arteries. In a second step, we compared these descriptors to human assessments of irregularity of the aneurysm dome.
Preliminary results are presented based on 134 aneurysm models and 15 raters (5 clinicians, 10 informed laymen). Univariate correlation revealed that curvature-based metrics predict most accurately the human assessment of irregularity (rank correlation: ρ=0.86, p<0.001), but also other geometry indices like aneurysm size (ρ=0.78, p<0.001) or writhe-based indices (ρ=0.75, p<0.001) are linked with overall perceived morphological complexity.
This study was performed within the scope of the AneuX project, funded by SystemsX.ch, and received support by SNSF NCCR Kidney.CH.