Critical area detection in cerebral aneurysm biomechanical models assisted by dimensionality reduction techniques

FRANCESCO CAMUSSONI; NICOLÁS MUZI; DANIEL MILLÁN; LUIS G. MOYANO

Buenos Aires

Simposio; 50 Jornada Argentina de Informática. Simposio Argentino de Ciencia de Datos.; 2021

49 Jornada Argentina de Informática. Simposio Argentino de Ciencia de Datos.

Intracranial aneurysms, along with other cerebrovascular accidents, is the third death cause in Argentina. The probability of rupure during a surgical intervention due to internal applied pressure is nonnegible, and surgeons often need to make decisions with scarse information. Here, we developed a number of finite element models, describing real aneuryms obtained from the public Aneurisk database, by means of Kirchhoff-Love thin-lamina theory. A series of numerical experiments were performed simulating a concentrated pressure applied to selected loading points in the wall of the aneurysms, producing a number of biomechanical features for every point in the model. The experiments encompassed several loading points at different locations of the aneurysm model. To better analyze these data, dimensionality reduction techniques were applied to the generated biomechanical features. Our results indicate that these techniques are an efficient method to capture sensitive areas where, for instance, buckling effects associated with large maximum curvatures and large maximum and mean displacement ratios are to be found, suggesting that they are able to pinpoint areas with high rupture risk in the event of excessive pressure. This ongoing work aims at providing a decision making support information for surgical intereventions of cerebral aneurysms.