Real time artificial intelligence assisted carotid artery stenting: a preliminary experience

Area of Science:

• Neurosurgery
• Medical Devices
• Artificial Intelligence

Background:

• Neurointerventional procedures require simultaneous monitoring of multiple devices on various screens, increasing the risk of oversights and complications.
• Artificial intelligence (AI) shows promise in enhancing device recognition and monitoring during fluoroscopic imaging in neurointerventional radiology.

Purpose of the Study:

• To evaluate the preliminary experience and efficacy of a real-time AI assistance software, Neuro-Vascular Assist, during carotid artery stenting procedures.
• To assess the safety and accuracy of AI-driven device monitoring in endovascular interventions.

Main Methods:

• A preliminary study involving six patients undergoing carotid artery stenting utilized the Neuro-Vascular Assist software.
• The software provided real-time tracking of wires, catheters, and embolic protection devices, alerting operators to deviations from predefined regions of interest.
• Efficacy, safety, and accuracy metrics were collected and analyzed.

Main Results:

• The Neuro-Vascular Assist software demonstrated effective and user-friendly operation.
• The system generated a mean of 21.0 notifications per procedure, with high precision (82%) and recall (94%).
• AI notifications prompted device repositioning in 23% of true positive alerts, with no software-related adverse events or periprocedural complications.

Conclusions:

• Neuro-Vascular Assist, a real-time AI assistance software, performed appropriately and may offer benefits in carotid artery stenting.
• Further large-scale studies are recommended to validate the utility and impact of this AI tool in neuroendovascular procedures.