Rapid tissue coverage of the fluoropolymer-coated watchman FLX pro: insights from automated deep-learning CT quantification in a high-bleeding-risk patient Authors: Qingsong Wang, Tao Chen, Jun Guo, Stefano Bordignon, Ugur Canpolat Published: 2026 Case presentation A 75-year-old man with paroxysmal atrial fibrillation presented with an acute ischaemic stroke, complicated by post-thrombolysis intracranial haemorrhage (ICH) and subsequent lower-limb necrosis necessitating amputation. Given the prohibitive coexistence of high thrombotic (CHA2DS2-VASc 7) and extreme haemorrhagic risks (HAS-BLED 3), standard full-dose oral anticoagulation was contraindicated. Percutaneous left atrial appendage closure (LAAC) was indicated as an alternative non-pharmacological strategy for stroke prevention. Imaging acquisition and implantation All cardiac computed tomography angiography (CCTA) scans in this study were acquired using a dual-source 256-slice CT scanner (SOMATOM Definition Flash, Siemens Healthineers, Forchheim, Germany) with retrospective electrocardiogram (ECG) gating. The tube voltage was auto-adjusted (70–150 kV), and 60 mL of non-ionic contrast medium was administered at 4.0–5.0 mL/s. Optimal cardiac phases (40% R-R interval for heart rate <70 beats per minute; 65% for ≥70) were reconstructed using an iterative algorithm (kernel Bv40) to minimize motion artifacts. Preprocedural 3D CCTA reconstruction guided the anatomical assessment (Figure 1A-B). A 31 mm Watchman FLX Pro (Boston Scientific, MA, USA) was successfully implanted (Figure 1C-D; Supplementary material online, Videos S1 and S2). This device features a novel fluoropolymer coating designed to enhance hemocompatibility. Due to the recent ICH, a deviation from the standard post-procedural antithrombotic protocol was clinically mandated. To balance the competing risks of device-related thrombosis (DRT) and fatal bleeding, the patient was prescribed a tailored, conservative regimen of low-dose Edoxaban (15 mg daily) postprocedure. Procedural imaging and 45-day automated CCTA quantification of the Watchman FLX Pro device. (A and B) Pre-procedural three-dimensional (3D) cardiac computed tomography angiography (CCTA) reconstructions evaluating the left atrial appendage (LAA) anatomy for precise device sizing. (C and D) Intra-procedural fluoroscopy demonstrating the successful deployment and stable positioning of the 31 mm Watchman FLX Pro device. (E) Follow-up CCTA at 45 days post-implantation confirming complete LAA sealing with no visible peri-device leak. (F and G) Advanced 3D volume rendering of the device’s atrial surface at 45 days, illustrating a smooth and uniformly distributed layer of hypoattenuated thickening (HAT). (H) Fully automated deep-learning (3D U-Net) quantitative analysis segmenting the HAT layer (distinguished by attenuation thresholds), revealing a rapid HAT Coverage Rate (HCR) of 91.7% over the exposed atrial surface of the device. (CCTA, cardiac computed tomography angiography; LAA, left atrial appendage; HAT, hypoattenuated thickening; HCR, HAT coverage rate). Advanced quantitative follow-up analysis Follow-up CCTA was performed at 45 days utilizing the aforementioned acquisition protocol to assess device healing. To ensure objective evaluation, we employed a validated, fully automated deep-learning approach using a 3D U-Net convolutional neural network model (Anythink β, Crealife Technology, Beijing). This artificial intelligence (AI) model automatically delineates the device boundary and segments hypoattenuated thickening (HAT). HAT was defined as tissue-like density on the atrial surface, distinguished from the contrast-enhanced blood pool by attenuation thresholds [< 100 Hounsfield units (HU) or a left atrial appendage (LAA) to left atrium (LA) HU ratio <0.25]. Healing was quantified by the HAT Coverage Rate (HCR), calculated as the segmented HAT area divided by the total exposed atrial surface area. Unlike manual visual assessment, this U-Net model has been shown to provide robust reproducibility. The 45-day CCTA demonstrated stable device positioning with no visible peri-device leak (Figure 1E). Automated 3D volume rendering visualized a layer of smooth, uniform HAT on the device surface (Figure 1F-G). Quantitative analysis revealed an HCR of 91.7% (Figure 1H). Clinically, after 3 months of low-dose Edoxaban, the patient was safely transitioned to dual antiplatelet therapy (DAPT). At the current 5-month follow-up, the patient remains free from DRT, recurrent stroke, and bleeding events. Discussion and conclusion This case utilizes AI-augmented CCTA quantification to illustrate rapid tissue coverage on the fluoropolymer-coated Watchman FLX Pro. Our team’s previous benchmark study on conventional uncoated Watchman devices showed that achieving ∼90% HAT coverage typically requires 6 to 12 months.1 Thus, reaching 91.7% coverage at merely 45 days suggests a highly favourable and rapid healing trajectory. We acknowledge that CCTA imaging alone cannot histologically distinguish mature endothelium from early fibrin deposition or inflammatory tissue. However, a pivotal pre-clinical histological study by Kramer et al. demonstrated that smooth HAT on FLX devices was strongly associated with a monolayer of typical endothelial cells over a fibrovascular matrix.2 This preclinical evidence suggests that the smooth HAT observed in our patient likely indicates a benign healing process rather than thrombus formation. The clinical relevance of this rapid tissue coverage is noteworthy, which provides a potential safety margin that may support the use of abbreviated or low-dose antithrombotic regimens in fragile patients with extreme bleeding risks. While the 5-month event-free follow-up is promising, extended large-scale clinical surveillance remains necessary to confirm if these favourable healing characteristics can consistently inform and optimize post-LAAC pharmacological protocols. Supplementary Material Contributor Information Qingsong Wang, Senior Department of Cardiology, Chinese PLA General Hospital, Fucheng Road 6, Beijing 100048, China; Department of Cardiology, No. 988 Hospital of PLA Joint Logistic Support Force, Zhengshang Road 602, Zhengzhou 450042, China. Tao Chen, Senior Department of Cardiology, Chinese PLA General Hospital, Fucheng Road 6, Beijing 100048, China. Jun Guo, Senior Department of Cardiology, Chinese PLA General Hospital, Fucheng Road 6, Beijing 100048, China. Supplementary material Supplementary material is available at European Heart Journal – Case Reports online. Author contributions Q.W., T.C., and J.G. cared for the patient. Q.W. drafted the manuscript and performed follow-up analysis. T.C. performed pre-procedural CT planning and image analysis, and performed the procedure with J.G. All authors approved the final manuscript and declared that no Artificial Intelligence (AI) or AI-assisted technologies were used in the creation of this manuscript. Written consent for publication was obtained from the patient. Qingsong Wang (Writing—review & editing [supporting]), Tao Chen (Writing—review & editing [supporting]), and Jun Guo (Writing—review & editing [supporting]). Funding This work was supported by the Capital’s Funds for Health Improvement and Research (CFH 2024–2-5071), the ‘Future Talent’ Training Program (MBRC0012025043), Henan Provincial Science and Technology Research Project (262102310310), and the Guidance Plan Project for Scientific and Technological Innovation (2025YLZDJH408). Data availability Supplementary material is available at European Heart Journal online. Data are openly available in a public repository. References Associated Data Supplementary Materials Data Availability Statement Supplementary material is available at European Heart Journal online. Data are openly available in a public repository.