Morphological descriptors as rupture indicators in middle cerebral artery aneurysms

Abstract

The rupture of intracranial aneurysms is associated to significant morbidity and mortality rates. Although the mechanisms triggering this event are still unclear, morphology is among the factors considered by interventional neuroradiologists to decide treatment. The aim of this work is to explore the potential of morphological descriptors as rupture risk predictors in middle cerebral artery aneurysms (MCA) and to provide the subset showing the best predictive capabilities. The set of evaluated descriptors include basic shape descriptors related to the aneurysm size, and most sophisticated ones such as the Zernike Moment Invariants. The population analyzed included 71 patients harboring 86 MCA aneurysms (64 unruptured vs. 22 ruptured). An existing image-based processing pipeline was used to extract such descriptors from Three-Dimensional Rotational Angiography (3DRA) images routinely acquired during standard clinical practice. Univariate and multivariate statistical analyses have shown that among the evaluated descriptors, Zernike moment invariants computed on the aneurysm and a small portion of the surrounding vessels, together with the non-sphericity index, provide the best predictive capabilities of aneurysm rupture.