KI-gestützte, automatisierte Aortenmorphometrie am Computertomogramm

In recent years applications from the field of artificial intelligence (AI) for fully automated aortic morphometry have increasingly been described. The aim of this review article is to analyze the study situation with respect to the possibilities and limitations of the currently described applications.

A literature search was carried out in PubMed. The study was carried out according to the PRISMA guidelines.

The search could identify 349 literature sources. After checking the abstracts 20 studies were included in the analysis. A multitude of AI-assisted applications have already been described in the literature, which enable a fully automated measurement of the diameter of the aorta with a high conformity to human reference measurements. The most frequently described applications include PRAEVAorta 2 (Nurea, Bordeaux, France), Augmented Radiology for Vascular Aneurysm (ARVA; Incepto, Paris, France) and AI Rad Companion Chest CT (Siemens Healthineers, Munich, Germany). Despite the positive results numerous segmentation errors have been described (up to 10%), which necessitate a critical assessment of the measurement results from human investigators. The quality of the data of the currently available studies is limited by possible bias due to selection. The evaluative comparison of the applications is currently only possible within limits due to the heterogeneous statistical work-up.

The fully automated measurement of the maximum aortic diameter and volume which is made possible by AI, offers great potential for the implementation in patient care and clinical research; however, at the present time the application is also coupled to relevant limitations, which must be taken into account in the utilization.