Left atrial appendage occlusion optimized with artificial intelligence-guided CT pre-planning and intra-procedural...

Left atrial appendage occlusion (LAAO) offers a safe alternative to oral anticoagulants for stroke prevention in patients with atrial fibrillation. This study compares the integration of preprocedural cardiac computed tomography (CT), artificial intelligence (AI) modeling, and intracardiac echocardiography (ICE) to standard transesophageal echocardiography (TEE)-guided LAAO.

This single-center retrospective cohort study from 2021 to 2024 compared TEE-guided LAAO to a method combining artificial intelligence-augmented CT preplanning and ICE. The primary outcome of the study was a composite of procedural success, incidence of device-related thrombus, and peri-device leaks ≥5 mm at 45 days. Secondary outcomes included procedural duration, contrast volume, device recaptures, and resizing.

The study included 143 patients: 46 in the CT/ICE arm and 97 in the TEE arm. Baseline characteristics were similar in the two groups. The number of males was statistically higher in the CT/ICE arm (76.1 % vs. 58.8 % p = 0.043). Procedural success was similar between groups (97.8 % vs. 95.9 %, p = 0.899). At 45 days, device-related thrombus (2.2 % vs. 0 %, p = 0.145) and peri-device leaks were similar (8.7 % vs. 4.1 %, p = 0.266) in both groups. The CT/ICE group had significantly fewer device recaptures (0.0 ± 0.1 vs. 0.2 ± 0.4, p = 0.001) and devices attempted (1.0 ± 0.2 vs. 1.2 ± 0.4, p = 0.002).

AI-augmented preprocedural cardiac CT planning combined with ICE has similar efficacy to TEE-guided LAAO. Due to the need for significantly fewer LAA devices and recaptures of these devices, there is the potential that AI augmented CT planning combined with ICE can reduce adverse events. The use of ICE and eliminating the need for general anesthesia and preprocedural TEE planning has the potential to improve procedural efficiency.