Cerebral emboli detection and autonomous neuromonitoring using robotic transcranial Doppler with artificial intelligence for transcatheter aortic valve replacement with and without embolic protection devices: a pilot study
Ammad A. Baig, Christopher Manion, Wasiq I. Khawar, Brianna M. Donnelly, Kunal P. Raygor, Ryan C Turner, David R. Holmes, Vijay Iyer, L. Nelson Hopkins, Jason M. Davies, Elad I. Levy, Adnan H. Siddiqui
- Year
- 2023
- Citations
- 5
Abstract
Background Periprocedural ischemic stroke remains a serious complication in patients undergoing transcatheter aortic valve replacement (TAVR). We used a novel robotic transcranial Doppler (TCD) system equipped with artificial intelligence (AI) for real-time continuous intraoperative neuromonitoring during TAVR to establish the safety and potential validity of this tool in detecting cerebral emboli, report the quantity and distribution of high intensity transient signals (HITS) with and without cerebral protection, and correlate HITS occurrence with various procedural steps. Methods Consecutive patients undergoing TAVR procedures during which the robotic system was used between October 2021 and May 2022 were prospectively enrolled in this pilot study. The robotic TCD system included autonomous adjustment of the TCD probes and AI-assisted post-processing of HITS and other cerebral flow parameters. Basic demographics and procedural details were recorded. Continuous variables were analyzed by a two-sample Mann–Whitney t-test and categorical variables by a χ 2 or Fisher test. Results Thirty-one patients were prospectively enrolled (mean age 79.9±7.6 years; 16 men (51.6%)). Mean aortic valve stenotic area was 0.7 cm 2 and mean aortic–ventricular gradient was 43 mmHg (IQR 31.5–50 mmHg). Cerebral protection was used in 16 cases (51.6%). Significantly fewer emboli were observed in the protection group than in the non-protection group (mean 470.38 vs 693.33; p=0.01). Emboli counts during valve positioning and implantation were significantly different in the protection and non-protection groups (mean 249.92 and 387.5, respectively; p=0.01). One (4%) transient ischemic attack occurred post-procedurally in the non-protection group. Conclusion We describe a novel real-time intraoperative neuromonitoring tool used in patients undergoing TAVR. Significantly fewer HITS were detected with protection. Valve positioning–implantation was the most significant stage for intraprocedural HITS.
Keywords
Related papers
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Fractional Differential Equations
Igor Podlubný
2025
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991
Genetic Programming: On the Programming of Computers by Means of Natural Selection
John R. Koza
1992