Rupture status discrimination in intracranial aneurysms using the centroid-radii model.

Abstract

Intracranial aneurysms are localized, abnormal arterial dilatations with a variable risk of rupture, leading to medical conditions associated with high morbidity and mortality. Predicting their risk of rupture, especially for incidental asymptomatic aneurysms, is a challenging task. The size of the aneurysm sac is traditionally used to assess the risk, but shape analysis has emerged as a promising differentiator of rupture likelihood. The centroid-radii model (CRM) is introduced here to describe both the size and the shape of the aneurysms, and determine rupture status. The entropy of CRM is proposed as an aneurysm descriptor which is easy to compute, robust to noise and segmentation, and accurate in rupture status discrimination. Analysis is performed on 154 patient-derived saccular aneurysms. The aneurysms are further classified as sidewall and bifurcation, and the shape analysis is performed separately on the two subtypes. Using the entropy of CRM resulted in 80.3% and 70.5% classification accuracy of status rupture in sidewall and bifurcation aneurysms, respectively. When compared to the accuracy of some commonly used size and shape indexes, the entropy of the CRM proved to be a more accurate single index associated with rupture in intracranial aneurysms, for both sidewall and bifurcation subtypes.