Publication type: Conference poster
Type of review: Peer review (abstract)
Title: Identification of clinically relevant characteristics of intracranial aneurysm morphology
Authors: Juchler, Norman
Schilling, Sabine
Bijlenga, Philippe
Rüfenacht, Daniel
Kurtcuoglu, Vartan
Hirsch, Sven
et. al: No
Conference details: 1. Digital Health Lab Day (Life in Numbers 5), Wädenswil, 3. October 2019
Issue Date: 2019
Publisher / Ed. Institution: ZHAW Zürcher Hochschule für Angewandte Wissenschaften
Language: English
Subjects: Intracranial aneurysms; Morphology
Subject (DDC): 616.8: Neurology, diseases of nervous system
Abstract: Intracranial aneurysms are focal deformations of larger cerebral arteries that occur in 2-5% of the population. Although they remain quiescent most of the time, aneurysms may rupture at a rate of about 1% per year, leading to subarachnoid hemorrhage with potentially catastrophic effects on the patient. It is exceedingly challenging to predict the clinical fate of intracranial aneurysms. Currently, physicians associate shape irregularity with vessel wall instability. However, there is no consensus on which shape features reliably predict aneurysm rupture. Here we present two approaches that aim to eliminate the subjectivity of rater assessment. In a first approach, we have implemented a semi-automated classification pipeline to predict the rupture status using morphometric parameters. These parameters were computed from 3D geometries of intracranial aneurysms obtained from 3D rotational angiographies. The main objective of this first study was to identify morphometric parameters that efficiently encode the disease status and understand how well morphology predicts disease status in general. In the second study, we followed a psychometric approach to better understand how human raters assess aneurysm morphology. We acquired rating data from 39 clinical experts and informed laypersons on perceived irregularity and the presence of 5 different morphological attributes (presence of a rough surface, of blebs, lobules, asymmetry and a complex parent vasculature). We related this data to clinically relevant parameters using regression analysis and binary classification. Our investigations confirmed that aneurysm morphology provides significant information about the disease. For example, a logistic regression model based on perceived irregularity alone is able to discriminate relatively well ruptured from unruptured aneurysms (AUC=0.81±0.04). Extending that model by aneurysm location increased the AUC significantly to 0.87±0.08, suggesting that morphology varies with location. Morphometric parameters well correlating to perceived irregularity (e.g. non-sphericity NSI, or total Gaussian curvature GLN) likewise predict the disease status well, but to a lesser extent. More specific parameters are required that are able to encode relevant morphological structure such as blebs/lobules or asymmetry.
Fulltext version: Published version
License (according to publishing contract): Licence according to publishing contract
Departement: Life Sciences and Facility Management
Organisational Unit: Institute of Computational Life Sciences (ICLS)
Published as part of the ZHAW project: AneuX
Appears in collections:Publikationen Life Sciences und Facility Management

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.