REGISTRY on the Implementation of Artificial Intelligence in the Automatic Analysis of Vascular Network Segmentation

REGISTRY on the Implementation of Artificial Intelligence in the Automatic Analysis of Vascular Network Segmentation

The team hypothesizes that fully automatic analysis of AAA could provide increased performance (decreased duration of segmentation with increased reproducibility and decreased inter and intraobserver variability) to detect aortic aneurysmal sac enlargement (volumes and diameters) and predict the risk of complications during the procedure and during follow-up (MAE, MACE, MALE, Stroke) compared to standard methods of measurement relying on approximate maximum sac diameter.

An abdominal aortic aneurysm (AAA) is an abnormal dilation of the abdominal aortic wall. The most catastrophic consequence of AAA increase evolution is aortic rupture, which still results in high morbidity and mortality. Accurate measurement of AAAs is necessary to predict the risk of rupture during evolution, to take adequate decisions to treat or not the patient as well as detect general and technical risk factors and to follow aneurysm sac behavior after endovascular aortic repair (EVAR).

Despite the widespread use of diameter measurements in clinical trials and its ease of ascertainment in clinical practice for monitoring of AAAs; clinical decision-making regarding the timing of aneurysm repair and even surveillance of sac expansion after EVAR, several studies have concluded that the diameter may not be reliable as a rupture risk criterion and that it should be replaced by more specifics criteria.

Volumetric assessment of the aneurysm is bound to be a better predictor of AAA expansion and risk of rupture. Recently, there has been considerable progress in segmentation software, allowing a semi-automatic calculation of accurate volumes from CTAs. However, despite ample evidence, volumetry has largely remained in the research domain and is still not carried out in most institutions. The major reason is that segmentation methods are time-consuming, they do not allow co-registration of interval studies and they require dedicated software and skilled technicians, which may be difficult to organize.

Innovative software PREAVAorta of Nurea company, using artificial intelligence with deep learning approaches, is able to reconstruct automatically the vascular structures from CT scans. As current solutions only reconstruct the lumen, Nurea's software also segments automatically aneurysms and associated thrombus. With this reconstruction, the software is able to provide diameters (and in particular maximum diameters) but also aneurysmal sac volume. It is the first solution providing automatic AAA volume and comparative evolution during follow-up. In addition, the software provides distances between anatomic points, calcification volume and measurement evolutions between different time points. The software also automatically detects and quantifies calcifications and stenosis on peripheral arteries, which is currently evaluated on the same or independent CT scans for predictive factor analysis, especially with regard to stroke risk for carotid stenosis and accesses, technical difficulties and predictors or MACE (Major Adverse Coronary Event) or MALE (Major Adverse Limb Event) for iliac and femoral arteries.

Inclusion Criteria:

• Male or female ≥18 years of age

• AAA treated with EVAR between

  • ≥50mm
  • or ≥5mm increase in 6 months

• The subject is legally competent, has been informed of the nature, the scope, and the relevance of the study and has not manifested an opposition to participate.

• Available for set-up and proper implementation of follow-up visits (standard of care) throughout the duration of the registry

Exclusion Criteria:

• Advancing Alzheimer's disease or socially dependent patient
• Opposition to the use of their data for this research
• Life expectancy supposed to be inferior to 2 years
• Pregnant woman