Early clinical experiences with AI-based EVAR planning using the Endoleak Risk Index support its value for...

Area of Science:

• Vascular Surgery
• Medical Imaging
• Artificial Intelligence

Background:

• Endovascular aortic repair (EVAR) is a common procedure for infrarenal aortic aneurysms.
• Type 1a endoleaks (EL1a) remain a significant complication, necessitating careful preoperative planning.
• The Endoleak Risk Index (ERI) is an artificial intelligence (AI)-based tool developed to predict EL1a risk.

Purpose of the Study:

• To evaluate the initial experience with the AI-based ERI in planning infrarenal EVAR.
• To assess the impact of ERI on clinical decision-making during EVAR planning.
• To investigate the predictive accuracy of ERI for EL1a in a real-world clinical setting.

Main Methods:

• A single-center study involving two groups of patients undergoing EVAR.
• Group 1: Retrospective cohort (n=10) with ERI calculated from preoperative CT angiography and compared to outcomes.
• Group 2: Prospective cohort (n=10) with AI-based simulations including ERI calculation for endograft sizing and EL1a risk prediction; ERI's influence on decisions was assessed.

Main Results:

• In Group 1, six of 10 patients had elevated ERI, and four experienced EL1a. Patients with low ERI remained endoleak-free.
• In Group 2, ERI influenced treatment decisions in three patients, including one change in endograft size and one case deemed unsuitable for EVAR.
• No EL1a occurred in Group 2 during a median 3-month follow-up.

Conclusions:

• The AI-based ERI shows potential value in EVAR planning, even in less complex cases.
• ERI calculation can aid surgeons in decision-making and potentially enhance patient safety.
• Further validation with larger datasets and technological advancements may solidify AI's role in predicting EL1a and optimizing EVAR strategies.