Next Article in Journal
Research on Heat Transfer Coefficient Prediction of Printed Circuit Plate Heat Exchanger Based on Deep Learning
Previous Article in Journal
Design and Optimization of an FPGA-Based Infrared Dim Small Target Detection Network Under a Sky Cloud Background
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 15
Issue 9
10.3390/app15094633
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Study of Atrial Fibrillation and Stroke Based on Geometrical and Hemodynamic Characteristics: A Review

by
Xiaoyu Liu
1,2 and
Qi Gao
1,2,*
1
Institute of Fluid Engineering, School of Aeronautics and Astronautics, Zhejiang University, Hangzhou 310007, China
2
State Key Laboratory of Transvascular Implantation Devices, Zhejiang University, Hangzhou 310007, China
*
Author to whom correspondence should be addressed.
Appl. Sci. 2025, 15(9), 4633; https://doi.org/10.3390/app15094633
Submission received: 19 March 2025 / Revised: 8 April 2025 / Accepted: 17 April 2025 / Published: 22 April 2025
Browse Figures
Versions Notes

Abstract

:
The CHA2DS2-VASc score is the most widely used and recognized method for stroke risk stratification in atrial fibrillation (AF) patients. However, some patients with low scores still experience strokes. Given that 90% of cardiogenic strokes are caused by thrombus in the left atrial appendage (LAA), it is essential to incorporate hemodynamic and geometric features of the LAA into existing risk stratification models. This review first evaluates current stroke and bleeding risk stratification strategies, then analyzes the geometric and hemodynamic parameters within the left atrium and LAA, and finally compares the methods and techniques available for acquiring these parameters. Through these retrospective analyses, insights and recommendations for the management of AF patients and stroke prevention are provided. Outlooks on future research directions, such as the exploration of the mechanism of thrombus detachment, are discussed.

1. Introduction

Atrial fibrillation (AF) is the most prevalent cardiac arrhythmia worldwide. As of 2019, approximately 59 million people globally were affected by AF [1]. A Chinese study involving over 10 million participants estimated the lifetime risk of developing AF to be around 20% [2]. Over a 50-year follow-up, the age-standardized prevalence of AF increased four-fold [3].
Patients with AF experience rapid and irregular electrical activity in the atria, contributing to complications such as heart failure, myocardial infarction, and thromboembolism [4]. Thromboembolism occurs due to reduced atrial contractility, which slows blood flow within the heart [5]. If a thrombus dislodges, it can travel through the circulatory system to cerebral vessels, potentially causing an ischemic stroke [6]. The left atrial appendage (LAA) is the most common site for thrombus formation in AF patients, and these thrombi can lead to strokes if they enter the bloodstream [7]. The risk of stroke in AF patients is increased five-fold [8], with AF being the cause of 25% of strokes among the elderly [8]. Furthermore, strokes related to AF typically result in more severe clinical outcomes compared to other types. AF patients are twice as likely to become bedridden due to stroke complications and have a higher mortality risk [9,10].
The current management of AF patients is primarily based on assessing the risk of stroke and bleeding [11,12]. However, these assessments rely solely on demographic and clinical data. Numerous studies [7,13,14,15] have demonstrated that including hemodynamic and geometrical parameters in stroke risk assessment methods can significantly enhance their accuracy. The study of hemodynamic mechanisms in AF patients who experience strokes should focus on the processes of thrombus formation and detachment. Despite its importance, there is currently a lack of research specifically analyzing thrombus detachment. Besides medical imaging techniques, existing research methodologies include both numerical and experimental approaches. Numerical methods comprise traditional computational fluid dynamics (CFD) [16,17,18] as well as advanced artificial intelligence (AI) [19,20,21] techniques. This work provides a comprehensive overview of the current status, fundamental principles, and related technologies in this field.

2. Literature Search

A literature search was performed on PubMed, Embase, CENTRAL (Cochrane Library), and Google Scholar, up to September 2024. For Section 3, which discusses current risk stratification strategies, the search terms were “specific risk stratification,” “meta-analysis”, and “stroke”. This yielded 17 meta-analyses and systematic reviews related to the stroke and bleeding risk stratifications detailed in Table 1 and Table 2, respectively (see Appendix A for details). For Section 4.2, which discusses stroke-related hemodynamic indexes, the search terms were “left atrial appendage”, “hemodynamic indexes”, and “stroke risk”. The screening process involved title review, abstract review, and finally, full-text review to identify novel LAA hemodynamic indexes associated with stroke, as summarized in Table 3.

3. Current Risk Stratification Methods

Risk stratification for patients with AF involves assessing the likelihood of complications such as stroke or bleeding. Anticoagulant medications, like warfarin, can prevent or treat approximately two-thirds of strokes and thrombotic events related to AF. However, these medications also increase the risk of bleeding, which can have a high mortality rate and lifelong implications [22,23]. Thus, identifying individuals at low risk is crucial to avoid unnecessary anticoagulant therapy. Healthcare providers must balance bleeding and stroke risks by selecting the most appropriate anticoagulants and dosages, and continually adjusting treatment as patient risk factors evolve. The CHA2DS2-VASc [42] score is the most widely used tool for assessing stroke risk, followed by the CHADS2 score [43], ABC score [44], and Framingham score [45]. For evaluating bleeding risk, the HAS-BLED score [46] is most commonly used, along with the ATRIA score [47], ABC bleeding score [48], and ORBIT score [49].
Four stroke risk stratifications are listed in Table 1 and the four bleeding risk stratifications are listed in Table 2. The c-statistic was the most commonly used evaluation index in these meta-analyses, though some studies also used risk ratios or odds ratios. For consistency, the c-statistic was used as the evaluation metric for the different risk stratification methods in this section. Table 1 and Table 2 summarize each risk stratification’s risk factors, c-statistics from the meta-analyses, and corresponding references. Among them, the c-statistic value is the maximum, minimum, and average of all c-critic values that appear in Table A1. A c-statistic closer to 1 indicates better predictive performance.
From Table 1 and Table 2, it is clear that current risk stratification tools for stroke and bleeding take into account factors such as patient characteristics, comorbidities, and biomarkers. Among the various stroke scoring methods, the CHA2DS2-VASc score encompasses a wider range of complications and medical history. Notably, the ABC score is the only tool that incorporates biomarkers. In contrast to stroke risk scores, bleeding risk scores also consider renal and liver function, as well as anemia. The c-statistics of various risk stratification tools show minimal variation, indicating poor to moderate predictive performance (c-statistic < 0.7). While the CHADS2 score seems to offer the best evaluative performance among them, several studies suggest that its ability to identify low-risk populations is less effective compared to the CHA2DS2-VASc score [50,51,52]. The CHA2DS2-VASc score remains the most commonly used for stroke risk stratification due to its simplicity. However, many researchers describe its predictive power as “modest” [28] and “inferior” [25]. On the one hand, numerous meta-analyses highlight the high heterogeneity of the CHA2DS2-VASc score across different studies [24,26,53]. On the other hand, a significant percentage of patients with low CHA2DS2-VASc scores are found to have LAA thrombus (5.2% in [52]). Some cohort studies and multivariable analyses even suggest no positive correlation between the CHA2DS2-VASc score and LAA thrombus presence, or no significant score differences between thrombus and control groups [54]. Additionally, the risk factors like sex, hypertension, heart failure, prior stroke, and diabetes as predictors of LAA thrombus have been questioned [55].
The relatively low accuracy of the CHA2DS2-VASc score could be attributed to its exclusive reliance on demographic and clinical data, lacking insights from modern imaging technology and hemodynamic analysis. Over a century ago, German pathologist Virchow identified three critical factors influencing thrombosis, known as Virchow’s triad: blood flow stasis, endothelial injury, and a hypercoagulable state [55]. This theory is still widely accepted today, emphasizing the significance of examining how hemodynamic and geometric parameters influence thrombus formation in the LAA and the associated risk of stroke (see Figure 1). Numerous studies employing imaging and CFD have shown more precise thrombotic risk predictions than the CHA2DS2-VASc score.

4. Relationship Between LAA and Stroke

The LAA is a protruding structure on the wall of the left atrium (LA) that plays a critical physiological role in regulating pressure within the LA [19]. However, due to its protruding anatomy and complex trabecular muscular structure, 70% to 90% of thrombus formation in non-valvular AF patients occurs in the LAA [56]. While biochemical and genetic factors are important, the geometric and hemodynamic characteristics of the LAA are the most direct contributors to this issue. This section will provide a thorough review of the geometric and hemodynamic features of the LAA to explore their potential relationship with stroke.

4.1. Geometric Characteristics of Left Atrial Appendage

The geometric features of the LAA can be described in terms of shape and dimensions. Wang et al. [57] identified four distinct shapes for the LAA: cactus, chicken wing, windsock, and cauliflower. Dimensions involve the quantitative analysis of specific geometric parameters of the LAA, such as volume, surface area, orifice diameter, and the number of lobes. The shape and dimensions of the LAA can provide complementary insights; for instance, the cactus shape typically exhibits a smaller orifice diameter and shorter central length, while the cauliflower shape often features a greater number of lobes [13,15,58].
The shape of the LAA provides an intuitive geometric description. In 2012, Di Biase et al. [13] proposed that the chicken-wing-shaped LAA is less likely to result in stroke or transient ischemic attack (TIA), while the cauliflower shape is most strongly associated with strokes. This concept has gained widespread acceptance in clinical practice, leading to the classification of the chicken wing as the “safest” LAA shape [59]. A growing body of evidence links the cauliflower-shaped LAA to a higher incidence of thrombus formation [6,58,60]. Additionally, a study conducted in 2015 [61] indicated that the windsock-shaped LAA, which has the largest average volume among the four types, is also more prone to strokes. However, other studies report no association between LAA shape and stroke risk [37,62]. This inconsistency underscores the need for further investigation. Furthermore, many studies recognize the challenges in accurately categorizing LAA shapes into these four types [63,64]; subtle features can be difficult to distinguish in clinical evaluations, particularly between cactus and windsock shapes [64]. Moreover, most studies have not definitively proven that thrombi in stroke events originate from the LAA, indicating the necessity for quantitative validation and confirmation.
In contrast, LAA dimension offers a more quantitative description, providing greater objectivity and reproducibility than shape classifications. Several consistent findings have emerged: an increased LAA volume, more lobes, greater centerline length, and larger LA diameter or volume correlate with a higher risk of thrombosis [6,33,38,61,62,65,66,67]. These dimensional features are thought to promote blood stagnation within the LAA. However, there is some debate over the influence of certain parameters on thrombotic risk, such as the LAA orifice [62,66]. Fang et al. [15] suggested that this debate arises from differing definitions of the orifice. Additionally, aging and chronic AF can lead to remodeling of the LA. Słodowska et al. [64] anatomically examined 200 cadaveric hearts and found that older adults often exhibit larger LAA volumes, longer lengths, and greater orifice areas. These age-related LAA changes potentially increase the amount of electrical activity and trigger atrial tachycardias. Furthermore, the relative position of the pulmonary veins to the LAA is also related to stroke risk; for instance, Nedios et al. [68] discovered that the higher starting points of the upper and lower edges of the LAA correlate with an increased risk of thromboembolism. Furthermore, Paliwal et al. [37] found that the area of the left inferior pulmonary vein in the stroke group was significantly larger than in the non-stroke group.
Compared to LAA shape, LAA dimension appears more suitable for guiding stroke risk stratification. Yaghi et al. [63] compared traditional methods (classifying the chicken wing shape as low risk) with the LAA folding angle to identify patient stroke risk, demonstrating the advantages of dimensional analysis. Additionally, Lei et al. [69] introduced the fractal dimension (FD) to describe the surface complexity of the LAA, finding that incorporating FD into the CHA2DS2-VASc score significantly enhanced stroke risk assessment performance (AUC = 0.8479 vs. 0.6958). Likewise, Yan et al. [70], Zhang and Yuan [71], Zhou et al. [72], and Chen et al. [73] included the LA diameter in the CHA2DS2-VASc score, which notably improved its performance as well (AUC = [0.71, 0.88]). These studies demonstrate that quantitatively incorporating LAA dimensions into traditional risk stratification enhances objectivity and improves repeatability. This approach is expected to guide future advancements in stroke risk stratification.

4.2. Hemodynamic Characteristics of Left Atrial Appendage

The process of thrombosis involves four key stages: vascular injury, platelet activation and aggregation, the coagulation cascade, and thrombus stabilization [74,75]. Initially, following vascular injury, platelets rapidly accumulate at the damage site and become activated, releasing chemicals that encourage the recruitment of additional platelets. Next, the activation of clotting factors through a cascade results in the formation of a fibrin mesh, which further strengthens the thrombus. Finally, the thrombus contracts and stabilizes over time, forming a robust structure that prevents blood flow [5,76]. Virchow’s triad is closely associated with this process [74,77]. Endothelial injury initiates the entire thrombus formation process. Blood flow stasis prolongs the retention of platelets and clotting factors at the site of injury, promoting their activation. In a hypercoagulable state, due to increased clotting factors, the coagulation cascade is more easily triggered. Thus, analyzing the flow patterns in the LA/LAA becomes essential.
Hemodynamics is closely linked to thrombus formation. When blood flow slows or eddies form, platelets more readily come into contact with the vessel wall and aggregate at sites of injury [7,16,78]. Low wall shear stress (WSS) can lead to endothelial dysfunction, while frequent changes in WSS direction are likely to trigger the activation of platelets, especially in areas with disturbed flow, such as bends or bifurcations [18,37,79]. In AF patients, irregular electrical signals reduce the contractility of the LA and decrease blood flow velocity. The peak flow velocity within the LAA is often less than 20 cm/s, promoting the accumulation and interaction of pro-coagulant factors, platelets, and red blood cells in this region [74,80,81]. Research shows that just 15 min of AF can increase thrombin levels, leading to a hypercoagulable state in the blood [82,83,84]. Several factors, including prolonged hemodynamic changes, atrial dilation, and fibrosis, can cause endothelial damage. This endothelial damage exposes matrix components, which enhance platelet adhesion and activation [85,86]. These factors align with Virchow’s triad, explaining the elevated risk of thrombosis in AF patients.
This review analyzes the hemodynamic mechanisms underlying Virchow’s triad, focusing on indicators related to the risk of stroke or thrombosis associated with the LAA. In Table 3, velocity stands out as the most direct and fundamental metric, measurable through both invasive and non-invasive imaging techniques. Research predominantly emphasizes mean LAA velocity, peak LAA velocity (both during emptying and filling phases), and peak LAA emptying velocity [7,15,78]. Nearly all studies acknowledge a substantial correlation between low flow velocity and the occurrence of AF, thrombosis, and stroke events.
The LAA ejection fraction is generally assessed using transesophageal echocardiography (TEE) [35,36]. The other indicators mentioned in Table 3 require simulation calculations based on geometric models. While attempts have been made to establish threshold values for these indicators to guide anticoagulant therapy, few studies provide such statistical thresholds.
Moreover, increased left atrial pressure (LAP) is observed in AF patients and might serve as an independent predictor for AF recurrence [19,87,88]. Nevertheless, there is a lack of studies directly investigating the relationship between elevated LAP and the risks of stroke or thrombosis, highlighting an area for further exploration.

5. Diagnostic Methods and Techniques

Stroke prevention is a primary focus in the study of the LAA and AF management. As discussed earlier, integrating the geometrical and hemodynamic characteristics of the LAA and LA into the existing CHA2DS2-VASc score may enhance stroke prediction by more effectively distinguishing between high- and low-risk patients. This section explores various methods for obtaining hemodynamic and geometric parameters, including both invasive and non-invasive imaging modalities, as well as computer-aided diagnostic methods.

5.1. Invasive Imaging Modalities

5.1.1. Transesophageal Echocardiography

TEE utilizes high-frequency sound waves to capture cardiac images via the esophagus and is widely used to detect thrombi and measure hemodynamic parameters (see Figure 2A,B). It is considered cost-effective and efficient for obtaining images, making it the “gold standard” for detecting LAA thrombi [89,90]. Studies have shown that TEE has a sensitivity and specificity approaching 100% for assessing LA thrombi [89]. Color Flow Doppler imaging aids in evaluating the LAA and identifying thrombi, while Tissue Doppler imaging assesses the myocardial function of the LAA region, providing additional information for thrombus risk stratification [91]. TEE’s high temporal resolution allows for continuous hemodynamic parameter curves throughout the cardiac cycle [92]. It also guides and evaluates LAA occlusion procedures by providing real-time imaging without the need for contrast agents or ionizing radiation [93], further establishing it as the “gold standard” for peri-procedural imaging during LAA occlusion procedures [94].
However, as an invasive procedure (see Figure 2C), TEE carries potential, albeit rare, life-threatening complications [98,99]. Traditional TEE may also struggle to accurately measure the hemodynamic parameters in thrombus regions due to its limitations in differentiating between muscle and thrombus, as well as in detecting thrombi obscured by anatomical protrusions [99]. To enhance the imaging quality of TEE, several strategies have been proposed. For instance, ultrasound contrast agents can serve as effective ultrasound reflectors, increasing blood echogenicity and minimizing imaging artifacts [100]. Additionally, a computer-aided diagnostic algorithm utilizing artificial neural networks has been developed to enhance the accuracy of TEE in diagnosing LA/LAA thrombi in patients with AF [101].

5.1.2. Intracardiac Echocardiography (ICE)

ICE involves inserting an ultrasound probe directly into the heart to obtain high-resolution cardiac images. It is a feasible and safe alternative to TEE for imaging during LAA occlusion procedures, potentially reducing the need for general anesthesia and its associated risks [102]. ICE is beneficial for assessing the geometry of the LAA and guiding device placement. However, achieving high-resolution images of the LAA with ICE remains challenging, even for experienced operators [103]. Additionally, ICE is generally more expensive and lacks multiplane imaging capabilities [104].

5.2. Non-Invasive Imaging Modalities

5.2.1. Transthoracic Echocardiography (TTE)

TTE is a non-invasive cardiac imaging technique that utilizes sound waves captured by an ultrasound probe placed on the patient’s chest to produce cardiac images, allowing for the assessment of heart structure and blood flow. Due to its non-invasive nature and lower cost, TTE is often the initial imaging modality used for AF patients to detect any structural abnormalities related to the condition [105]. However, it is widely recognized that TTE is less effective than TEE in detecting structural abnormalities and intracardiac thrombi [106,107]. Some studies suggest that the combined use of 2D-TTE and 3D-TTE can achieve an accuracy comparable to that of TEE in assessing thrombi within the LA and LAA [108].

5.2.2. Cardiac Computed Tomography (CCT)

CCT is a diagnostic imaging technique that utilizes X-rays to create detailed cardiac images. This technology can generate 3D geometric data of the entire heart and reconstruct images across different cardiac phases and planes, facilitating accurate evaluation of the LAA anatomy (see Figure 2D). As a result, CCT is considered the “gold standard” for visualizing the LAA and is widely employed for obtaining LAA morphological parameters [109]. However, CCT has limitations in distinguishing between thrombi and stagnant blood flow, which can lead to lower specificity in thrombus detection [110]. Additional disadvantages of CCT include significantly higher radiation exposure, the use of iodine-based contrast agents, and lower temporal resolution compared to TEE [111].

5.2.3. Cardiac Magnetic Resonance Imaging (MRI)

Cardiac MRI is a diagnostic technique that utilizes strong magnetic fields and radio waves to generate cardiac images. This method is effective in distinguishing between fresh thrombi, which exhibit increased signal intensity, and older thrombi that display decreased signal intensity, making it more effective than TEE in thrombus detection [112,113]. Additionally, cardiac MRI can measure flow velocity in the LAA and has shown good correlation with TEE [114]. However, it may overestimate the area of thrombi [115]. Its limitations include lower spatial resolution, the necessity for gadolinium-based contrast agents, and incompatibility with patients who have implanted cardiac devices [116].
With the ongoing development of phase-contrast (PC) MRI, time-resolved 3D PC MRI with velocity encoding in three directions (known as 4D flow MRI) has become widely adopted [117]. This technique allows comprehensive measurement of 3D hemodynamics in the heart, providing global coverage throughout the entire cardiac cycle. The resulting data, which include 3D spatial information integrated with time and three velocity directions, enable the calculation of various derived hemodynamic parameters, such as peak and mean flow velocity, stagnation areas, WSS, kinetic energy, and pressure gradients [7,78,118,119]. By analyzing LA stasis velocity and peak velocity, insights into the extent of blood stasis in the LAA of AF patients can be inferred [78,120]. Despite its advantages, 4D flow MRI remains a hot topic in cardiovascular imaging; however, the technique is hindered by limited temporal and spatial resolution, which can lead to the loss of flow information [117,121]. Wåhlin et al. [122] noted that, compared to 2D PC-MRI, the lower spatial resolution of 4D flow MRI tends to overestimate flow in smaller vessels due to partial volume effects.

5.3. Computer Aided Diagnosis

The various clinical imaging modalities discussed in the previous sections provide a direct and comprehensive characterization of the LA and LAA. However, these methods are often limited by factors such as spatial or temporal resolution and operational costs. With advancements in computer hardware and interdisciplinary collaboration, computer-assisted approaches have emerged as potential alternatives for assessing stroke risk in AF patients [14,16,19,20,39,123], simulating LAA occlusion procedures [124,125,126,127], and evaluating blood stasis within the LAA [17,41,128]. This section primarily reviews the applications of CFD and AI in the study of the LAA.

5.3.1. Computational Fluid Dynamics

CFD achieves high temporal and spatial resolution of flow field parameters by solving the Navier–Stokes (NS) equations over a user-defined fluid domain. A typical CFD simulation pipeline consists of reconstructing a 3D geometric model from medical images, generating a mesh, and setting the simulation parameters. CFD can be utilized to analyze the hemodynamic characteristics of the LA and LAA in AF patients [14,16,17,128], simulate pre- and post-operative conditions [18,127,129,130] (see Figure 3), investigate the mechanisms behind thrombus formation in the LAA [16,39,124], and identify new metrics for stroke risk assessment [14,16,18]. The indexes in the latter part of Table 3 were derived from CFD simulations. As a modern tool, CFD provides high temporal and spatial resolution, comprehensive coverage, and accuracy.
However, the limitations of CFD must also be acknowledged. Firstly, different CFD simulation studies often employ varying assumptions regarding models and computational settings, such as boundary conditions, initial conditions, and mesh resolution. These differences lack reliable clinical validation, complicating the assessment of their impact on simulation results [18,39,127]. Currently, there is no consensus on the most appropriate CFD simulation parameters and pipeline for the LA and LAA. Commonly used inlet boundary conditions typically include the velocities, pressures, and flow rates of the four pulmonary veins, often sourced from assumed values, clinical measurements, or flow balance calculations [127]. Similarly, typical outlet boundary conditions involve velocities, pressures, and flow rates at the mitral valve, which are also derived from fixed values or clinical measurements [127]. García Villalba et al. [39] compared the effects of rigid versus flexible wall assumptions, pulmonary vein boundary conditions, and laminar versus Large Eddy Simulation (LES) models on simulation outcomes, concluding that laminar flow and rigid wall assumptions are considered reliable.
Secondly, the high cost of CFD calculations poses a significant challenge to CFD’s development. Due to resource limitations, CFD studies are often restricted to small patient cohorts. This limitation, combined with a lack of reliable clinical validation, hampers the widespread application of CFD methods in clinical settings. Nonetheless, the rapid advancement of CFD technology remains notable; for instance, the HeartFlow software (Version: Heartflow One; Website: https://www.heartflow.com/) for assessing coronary artery disease has received approval from the U.S. Food and Drug Administration [132], demonstrating the potential for CFD tools in clinical applications.

5.3.2. Artificial Intelligence

AI is increasingly applied in clinical settings due to its strong performance in nonlinear high-dimensional analyses, such as image recognition and data regression [21,133,134]. Its applications in the context of the LA and LAA include extracting morphological information from medical images [133,135], calculating hemodynamic indices [19,20,21,136], and developing risk stratification or subtype classification based on specific indicators [134,137].
Figure 4 and Table 4 illustrate the current research pipeline regarding AI applications in LAA/LA studies. In Figure 4, the pipeline represented by blue arrows outlines the process of integrating hemodynamic and geometric parameters into stroke risk stratification without the use of AI technology. The blue arrow labeled a denotes the model reconstruction and segmentation from medical imaging using non-AI methods, such as threshold segmentation, region growing, and edge detection. The output of this step is the geometrical model (for b) or geometrical parameters (for d). The blue arrow b represents the use of CFD methods. The blue arrows c and d indicate the expert judgment of experienced researchers or medical professionals, highlighting that each of these processes incurs significant costs in computation, manpower, and time.
In contrast, the orange arrows represent three AI tasks (A/B/C) that can greatly simplify these processes. Task A involves utilizing AI techniques for reconstructing and segmenting models, as well as for identifying lesions in medical imaging. Task B refers to AI methods that predict hemodynamic parameters on the internal surface or wall of LA/LAA geometric models as an alternative to CFD. Task C pertains to AI technologies that directly diagnose stroke subtypes or assess stroke risk scores based on medical imaging. Currently, the clinical application of Task C often necessitates expert supervision of its results. However, with the continued advancement of AI explainability, algorithmic transparency, and generalization capabilities, AI holds the potential to become a significantly more efficient tool. This progress will foster a complementary partnership between AI and expert judgment.
Table 4 summarizes the references corresponding to Tasks A, B, and C. In this table, the most commonly used neural network (NN) related to imaging is U-Net (Tasks A/C), while the most frequently employed NN for flow field prediction is Graph-based Deep Learning (GDL) (Task B). U-Net is structured with a contracting path and an expanding path, which allows for the creation of highly detailed segmentation maps using a limited number of training samples. This is particularly important in the medical field, where labeled images are often scarce. Furthermore, due to its context-based learning approach, U-Net typically trains much faster than many other segmentation models [138,139].
On the other hand, GDL aggregates features from a given point along with those from its neighboring points to generate feature maps. This method significantly enhances the ability of traditional NNs to capture geometric features. Improvements in the interpretability of GDL can also assist clinicians in diagnosing diseases, such as identifying brain regions most relevant to specific tasks [140,141]. However, similar applications of GDL have not yet been explored in the context of LA/LAA tasks.
In addition to the aforementioned applications, AI technologies excel at the induction and extraction of multidimensional indicators or datasets. Xia et al. [142] proposed the H-BRBp disease diagnosis model, which combines expert knowledge in the modeling process with a data-driven approach in the training phase. This model demonstrates significant advantages in extracting strongly correlated indices related to lumbar spine diseases compared to traditional algorithms. Once the hemodynamic or geometric indicators of the LA/LAA are obtained, similar methods can be employed to extract key indicators and incorporate them into existing risk stratification frameworks.

6. Discussion

This review highlights the current challenge of insufficient accuracy in stroke risk stratification within clinical practice. It argues that the limitations of existing stratification methods stem from their over-reliance on demographic and clinical data, neglecting the valuable insights offered by modern imaging techniques and hemodynamic analysis. The review then explores potential avenues for improving stroke risk stratification, focusing on the shape, dimensions, and hemodynamic characteristics of the LAA. A comprehensive overview of clinical techniques and methods for acquiring these parameters, including advanced imaging technologies and computer-aided analysis, is provided.
Specifically, the review highlights the superior clinical applicability of quantitative dimensional assessment of the LAA compared to shape classification for stroke risk evaluation. Current clinical practice faces challenges in distinguishing subtle differences between the four classical LAA shapes, whereas dimensional measurements offer superior reproducibility and objectivity. Consistent findings indicate that increased LAA volume, higher lobe count, longer centerline length, and larger ostial diameter are all associated with a higher risk of thrombogenesis. However, further large-scale clinical-statistical analyses are needed to clarify the roles of factors such as LAA orifice characteristics and pulmonary vein positioning.
Furthermore, this review elucidates the relationship between Virchow’s triad and LAA hemodynamic parameters, summarizing current evidence on stroke-associated hemodynamic indicators (Table 3). Existing indicators primarily focus on flow velocity and wall shear stress derivatives, with limited attention paid to pressure. Future investigations should explore potential correlations between intracardiac pressure variations and stroke risk. Importantly, current research predominantly focuses on the mechanisms of thrombosis. However, the risk of stroke is significantly reduced if a thrombus remains attached. Therefore, future studies should also prioritize investigating the mechanisms underlying thrombus detachment and subsequent embolic events.
Translating these findings into clinical practice faces several obstacles. Accurate assessment of LAA geometry and hemodynamics requires advanced imaging modalities that may not be universally accessible. The subjective nature of medical image interpretation and diagnosis, influenced by factors such as clinician experience, knowledge base, and cognitive processes, can lead to inter-observer variability. Moreover, the computational burden of CFD simulations and the inherent risks and costs associated with invasive imaging techniques represent significant barriers to routine clinical implementation. Future research efforts should prioritize the development and validation of AI-driven solutions and non-invasive imaging technologies, specifically exploring the potential of GDL applications for pathological recognition and image segmentation.
This review acknowledges several limitations. First, methodological heterogeneity across the included studies, encompassing variations in research design, patient populations, and outcome measures, may introduce bias. Second, while this review focuses on geometric and hemodynamic factors, other crucial elements such as genetic predisposition and inflammatory markers are not explored in detail.

7. Conclusions

This review provides an overview of current risk stratification, examining the geometric and hemodynamic features of the LA/LAA related to stroke risk, as well as the clinical diagnostic methods and technologies available. It offers insights and recommendations for future stroke prevention and AF management. Future research on stroke risk could focus on thrombosis detachment, an area that has not yet been thoroughly studied. Integrating LA/LAA morphological and hemodynamic parameters related to thrombosis formation and detachment into existing risk stratifications could enhance the accuracy of risk evaluations. Furthermore, ongoing advancements in AI technologies may enable these tools to replace entire CFD workflows, leveraging expert knowledge or physical constraints. Such methods and techniques will help promote the further development of stroke risk stratification, facilitating the identification and management of medium- to high-risk AF patients.

Author Contributions

X.L.: Writing—Original Draft, Data analysis, Visualization. Q.G.: Conceptualization, Methodology, Writing—Review and Editing, Supervision, Funding acquisition. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant No. 12425208).

Conflicts of Interest

The authors declare that they have no competing financial or non-financial interests related to the work submitted for publication. The authors also confirm that there are no other personal relationships or circumstances that may have influenced the work presented in this paper.

Appendix A. Literature Review on Risk Stratification

A total of 17 meta-analyses and systematic review articles were retrieved, which are related to the four stroke risk stratifications in Table 1 and the four bleeding risk stratifications listed in Table 2. The relevant information has been summarized in Table A1.
Table A1. Studies on risk stratification meta-analysis and systematic review.
Table A1. Studies on risk stratification meta-analysis and systematic review.
StudyNumber of StudySample
Size		Risk Score CharacteristicsEvaluating Results
[51]12205,939CHADS2/CHA2DS2-VAScCHADS2: RR = 3.36 (2.93–3.85)
CHA2DS2-VASc: RR = 5.15(3.85–6.88)
[52]631,539CHADS2/CHA2DS2-VAScRR
[143]19714,672CHADS2Odds ratio
[53]10166,017CHA2DS2-VAScThe summary annual risk of stroke = 1.61%
[24]69845CHADS2/CHA2DS2-VAScCHADS2: DOR = 2.86 (1.79–4.55)
AUC = 0.6728
CHA2DS2-VASc: DOR = 2.8 (1.83–4.28)
AUC = 0.6655
[26]19846,748CHA2DS2-VAScC-statistic = 0.64–0.71
[27]6363,432CHA2DS2-VASc/ATRIACHA2DS2-VASc: C-statistic = 0.63
ATRIA: C-statistic = 0.66
[30]718,694ORBIT/HAS-BLEDORBIT: C-statistic = 0.65 (0.60–0.69)
HAS-BLED: C-statistic = 0.63 (0.60–0.67)
[22]50669,217ATRIA/ATRIA
bleeding/CHADS2/CHA2DS2-VASc/HAS-BLED/ORBIT		CHADS2: C-statistic = 0.64 (0.63–0.65) CHA2DS2-VASc: C-statistic = 0.62 (0.61–0.64)
HAS-BLED: C-statistic = 0.62 (0.58–0.66)
[144]9101,118HAS-BLED/ATRIA/ORBITodds ratio
[145]18321,888ABC-bleeding score/ATRIA/HAS-BLED/ORBIT/Odds ratio/sensitivity and specificity
analysis
[146]156223CHA2DS2-VAScOdds ratio/I2/sensitivity analysis
[23]3910,000+HAS-
BLED/ORBIT/ATRIA/CHA2DS2-VASc/CHADS2/ABC		HAS-BLED: 0.63 [0.61, 0.65]
ORBIT: 0.63 [0.60, 0.67]
ATRIA: 0.63 [0.60, 0.66]
CHADS2: 0.61 [0.57, 0.65]
CHA2DS2-VASc: 0.61 [0.57, 0.66]
ABC: 0.65 [0.58, 0.72]
[25]1106,627,101CHA2DS2-VASc/CHADS2/ATRIA/Framingham/The ABC stroke risk scoreCHA2DS2-VASc: 0.644 [0.635–0.653]
CHADS2: 0.658 (0.644–0.672)
ATRIA: 0.683 (0.658–0.708)
The ABC stroke risk score: 0.678 (0.658–0.697)
[28]19592,009CHA2DS2-VAScc-statistic = 0.66 (0.63–0.69)
[29]28639,450CHA2DS2-VAScc-statistic = 0.66 (0.62–0.70)
[31]17305,498HAS-BLED/ORBITHAS-BLED: 0.63 (0.60–0.66)
ORBIT: 0.61 (0.59–0.63)
[11]1212,510ABC stroke/CHA2DS2-VASc/ABC bleedingABC stroke: c-statistic = 0.67 (0.65–0.68)
CHA2DS2-VASc: c-statistic = 0.64 (0.60–0.67)
ABCbleeding: c-statistic = 0.66 (0.61–0.70)
[32]14–HAS-BLED/ORBIT/ABC bleedingHAS-BLED: 0.63 (0.61–0.65)
ORBIT: 0.65 (0.62–0.68)
ABC bleeding: 0.68 (0.61–0.75)

References

  1. Roth, G.A.; Mensah, G.A.; Johnson, C.O.; Addolorato, G.; Ammirati, E.; Baddour, L.M.; Barengo, N.C.; Beaton, A.Z.; Benjamin, E.J.; Benziger, C.P.; et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: Update from the GBD 2019 study. J. Am. Coll. Cardiol. 2020, 76, 2982–3021. [Google Scholar] [CrossRef] [PubMed]
  2. Guo, Y.; Tian, Y.; Wang, H.; Si, Q.; Wang, Y.; Lip, G.Y. Prevalence, incidence, and lifetime risk of atrial fibrillation in China: New insights into the global burden of atrial fibrillation. Chest 2015, 147, 109–119. [Google Scholar] [CrossRef]
  3. Schnabel, R.B.; Yin, X.; Gona, P.; Larson, M.G.; Beiser, A.S.; McManus, D.D.; Newton-Cheh, C.; Lubitz, S.A.; Magnani, J.W.; Ellinor, P.T.; et al. 50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the framingham heart study: A cohort study. Lancet 2015, 386, 154–162. [Google Scholar] [CrossRef] [PubMed]
  4. Bizhanov, K.A.; Abzaliyev, K.B.; Baimbetov, A.K.; Sarsenbayeva, A.B.; Lyan, E. Atrial fibrillation: Epidemiology, pathophysiology, and clinical complications (literature review). J. Cardiovasc. Electrophysiol. 2023, 34, 153–165. [Google Scholar] [CrossRef] [PubMed]
  5. Ding, W.Y.; Gupta, D.; Lip, G.Y.H. Atrial fibrillation and the prothrombotic state: Revisiting virchow’s triad in 2020. Heart 2020, 106, 1463–1468. [Google Scholar] [CrossRef]
  6. Lee, J.M.; Seo, J.; Uhm, J.; Kim, Y.J.; Lee, H.; Kim, J.; Sung, J.; Pak, H.; Lee, M.; Joung, B. Why Is Left Atrial Appendage Morphology Related to Strokes? An Analysis of the Flow Velocity and Orifice Size of the Left Atrial Appendage. J. Cardiovasc. Electrophysiol. 2015, 26, 922–927. [Google Scholar] [CrossRef]
  7. Garcia, J.; Sheitt, H.; Bristow, M.S.; Lydell, C.; Howarth, A.G.; Heydari, B.; Prato, F.S.; Drangova, M.; Thornhill, R.E.; Nery, P.; et al. Left atrial vortex size and velocity distributions by 4D flow MRI in patients with paroxysmal atrial fibrillation: Associations with age and CHA2DS2-VASc risk score. J. Magn. Reson. Imaging 2020, 51, 871–884. [Google Scholar] [CrossRef]
  8. Writing Group Members; Lloyd-Jones, D.; Adams, R.J.; Brown, T.M.; Carnethon, M.; Dai, S.; De Simone, G.; Ferguson, T.B.; Ford, E.; Furie, K.; et al. Heart disease and stroke statistics—2010 update: A report from the American Heart Association. Circulation 2010, 121, e46–e215. [Google Scholar]
  9. Lin, H.-J.; Wolf, P.A.; Kelly-Hayes, M.; Beiser, A.S.; Kase, C.S.; Benjamin, E.J.; D’Agostino, R.B. Stroke severity in atrial fibrillation. Stroke 1996, 27, 1760–1764. [Google Scholar] [CrossRef]
  10. Dulli, D.A.; Stanko, H.; Levine, R.L. Atrial fibrillation is associated with severe acute ischemic stroke. Neuroepidemiology 2003, 22, 118–123. [Google Scholar] [CrossRef] [PubMed]
  11. Ali, A.; Siddiqui, A.A.; Ali, M.; Shahid, I. Meta-analysis on performance of ABC and GARFIELD-AF compared to CHA2DS2-VASc and HAS-BLED in anticoagulated atrial fibrillation patients. Cardiovasc. Revascularization Med. 2024, 60, 74–81. [Google Scholar] [CrossRef]
  12. Zhu, W.; He, W.; Guo, L.; Wang, X.; Hong, K. The HAS-BLED score for predicting major bleeding risk in anticoagulated patients with atrial fibrillation: A systematic review and meta-analysis. Clin. Cardiol. 2015, 38, 555–561. [Google Scholar] [CrossRef]
  13. Di Biase, L.; Santangeli, P.; Anselmino, M.; Mohanty, P.; Salvetti, I.; Gili, S.; Horton, R.; Sanchez, J.E.; Bai, R.; Mohanty, S.; et al. Does the left atrial appendage morphology correlate with the risk of stroke in patients with atrial fibrillation? results from a multicenter study. J. Am. Coll. Cardiol. 2012, 60, 531–538. [Google Scholar] [CrossRef] [PubMed]
  14. Dueñas-Pamplona, J.; García, J.G.; Castro, F.; Muñoz-Paniagua, J.; Goicolea, J.; Sierra-Pallares, J. Morphing the left atrium geometry: A deeper insight into blood stasis within the left atrial appendage. Appl. Math. Model. 2022, 108, 27–45. [Google Scholar] [CrossRef]
  15. Fang, R.; Wang, Z.; Zhao, X.; Wang, J.; Li, Y.; Zhang, Y.; Chen, Q.; Wang, J.; Liu, Q.; Chen, M.; et al. Stroke risk evaluation for patients with atrial fibrillation: Insights from left atrial appendage with fluid-structure interaction analysis. Comput. Biol. Med. 2022, 148, 105897. [Google Scholar] [CrossRef]
  16. Vella, D.; Monteleone, A.; Musotto, G.; Bosi, G.M.; Burriesci, G. Effect of the alterations in contractility and morphology produced by atrial fibrillation on the thrombosis potential of the left atrial appendage. Front. Bioeng. Biotechnol. 2021, 9, 586041. [Google Scholar] [CrossRef]
  17. Vedula, V.; George, R.; Younes, L.; Mittal, R. Hemodynamics in the left atrium and its effect on ventricular flow patterns. J. Biomech. Eng. 2015, 137, 111003. [Google Scholar] [CrossRef]
  18. Dueñas-Pamplona, J.; García, J.G.; Sierra-Pallares, J.; Ferrera, C.; Agujetas, R.; López-Mínguez, J.R. A comprehensive comparison of various patient-specific CFD models of the left atrium for atrial fibrillation patients. Comput. Biol. Med. 2021, 133, 104423. [Google Scholar] [CrossRef]
  19. Liu, X.; Lin, H.; Liu, X.; Qian, J.; Cai, S.; Fan, H.; Gao, Q. LAFlowNet: A dynamic graph method for the prediction of velocity and pressure fields in left atrium and left atrial appendage. Eng. Appl. Artif. Intell. 2024, 136, 108896. [Google Scholar] [CrossRef]
  20. Gao, Q.; Lin, H.; Qian, J.; Liu, X.; Cai, S.; Li, H.; Fan, H.; Zheng, Z. A deep learning model for efficient end-to-end stratification of thrombotic risk in left atrial appendage. Eng. Appl. Artif. Intell. 2023, 126, 107187. [Google Scholar] [CrossRef]
  21. Ferez, X.M.; Mill, J.; Juhl, K.A.; Acebes, C.; Iriart, X.; Legghe, B.; Cochet, H.; De Backer, O.; Paulsen, R.R.; Camara, O. Deep learning framework for real-time estimation of in-silico thrombotic risk indices in the left atrial appendage. Front. Physiol. 2021, 12, 694945. [Google Scholar] [CrossRef]
  22. Proietti, M.; Farcomeni, A.; Romiti, G.F.; Di Rocco, A.; Placentino, F.; Diemberger, I.; Lip, G.Y.; Boriani, G. Association between clinical risk scores and mortality in atrial fibrillation: Systematic review and network meta-regression of 669,000 patients. Eur. J. Prev. Cardiol. 2020, 27, 633–644. [Google Scholar] [CrossRef]
  23. Gao, X.; Cai, X.; Yang, Y.; Zhou, Y.; Zhu, W. Diagnostic Accuracy of the HAS-BLED Bleeding Score in VKA- or DOAC-Treated Patients With Atrial Fibrillation: A Systematic Review and Meta-Analysis. Front. Cardiovasc. Med. 2021, 8, 757087. [Google Scholar] [CrossRef] [PubMed]
  24. Liu, F.D.; Shen, X.L.; Zhao, R.; Li, G.F.; Wu, Y.L.; Tao, X.X.; Wang, S.; Zhou, J.J.; Zheng, B.; Zhang, Q.T.; et al. Predictive role of CHADS2 and CHA2DS2-VASc scores on stroke and thromboembolism in patients without atrial fibrillation: A meta-analysis. Ann. Med. 2016, 48, 367–375. [Google Scholar] [CrossRef] [PubMed]
  25. van der Endt, V.H.W.; Milders, J.; de Vries, B.B.L.P.; Trines, S.A.; Groenwold, R.H.H.; Dekkers, O.M.; Trevisan, M.; Carrero, J.J.; van Diepen, M.; Dekker, F.W.; et al. Comprehensive comparison of stroke risk score performance: A systematic review and meta-analysis among 6 267 728 patients with atrial fibrillation. Europace 2022, 24, 1739–1753. [Google Scholar] [CrossRef] [PubMed]
  26. van Doorn, S.; Debray, T.P.A.; Kaasenbrood, F.; Hoes, A.W.; Rutten, F.H.; Moons, K.G.M.; Geersing, G. Predictive performance of the CHA2DS2-VASc rule in atrial fibrillation: A systematic review and meta-analysis. J. Thromb. Haemost. 2017, 15, 1065–1077. [Google Scholar] [CrossRef]
  27. Zhu, W.; Fu, L.; Ding, Y.; Huang, L.; Xu, Z.; Hu, J.; Hong, K. Meta-analysis of ATRIA versus CHA2DS2-VASc for predicting stroke and thromboembolism in patients with atrial fibrillation. Int. J. Cardiol. 2017, 227, 436–442. [Google Scholar] [CrossRef]
  28. Siddiqi, T.J.; Usman, M.S.; Shahid, I.; Ahmed, J.; Khan, S.U.; Ya’qoub, L.; Rihal, C.S.; Alkhouli, M. Utility of the CHA2DS2-VASc score for predicting ischaemic stroke in patients with or without atrial fibrillation: A systematic review and meta-analysis. Eur. J. Prev. Cardiol. 2022, 29, 625–631. [Google Scholar] [CrossRef]
  29. Basit, J.; Zaheer, Z.; Yasmin, F.; Ahmed, A.; Hamza, M.; Cheema, H.A.; Akbar, U.A.; Aamir, M.; Dhama, K.; Saeed, S.; et al. Abstract 17712: The Discrimination Ability of the CHA2DS2-VASc Score for Predicting Ischemic Stroke in Patients With and Without Atrial Fibrillation: An Updated Meta-Analysis. Circulation 2023, 148 (Suppl. S1), A17712. [Google Scholar] [CrossRef]
  30. Wang, C.; Yu, Y.; Zhu, W.; Yu, J.; Lip, G.Y.; Hong, K. Comparing the ORBIT and HAS-BLED bleeding risk scores in anticoagulated atrial fibrillation patients: A systematic review and meta-analysis. Oncotarget 2017, 8, 109703–109711. [Google Scholar] [CrossRef]
  31. Liu, X.; Wang, S.; He, W.; Guo, L. HAS-BLED vs. ORBIT scores in anticoagulated patients with atrial fibrillation: A systematic review and meta-analysis. Front. Cardiovasc. Med. 2023, 9, 1042763. [Google Scholar] [CrossRef]
  32. Zeng, S.; Zhu, W.; Guo, S.; Hong, H. Predictive performance of HAS-BLED, ORBIT, ABC, and DOAC scores for major bleeding in atrial fibrillation patients on DOACs. Eur. J. Intern. Med. 2024, 128, 131–133. [Google Scholar] [CrossRef] [PubMed]
  33. Pons, M.I.; Mill, J.; Fernandez-Quilez, A.; Olivares, A.L.; Silva, E.; de Potter, T.; Camara, O. Joint Analysis of Morphological Parameters and In Silico Haemodynamics of the Left Atrial Appendage for Thrombogenic Risk Assessment. J. Interv. Cardiol. 2022, 2022, 9125224. [Google Scholar] [CrossRef] [PubMed]
  34. Masci, A.; Barone, L.; Dedè, L.; Fedele, M.; Tomasi, C.; Quarteroni, A.; Corsi, C. The Impact of Left Atrium Appendage Morphology on Stroke Risk Assessment in Atrial Fibrillation: A Computational Fluid Dynamics Study. Front. Physiol. 2019, 9, 1938. [Google Scholar] [CrossRef]
  35. Park, H.-C.; Shin, J.; Ban, J.-E.; Choi, J.-I.; Park, S.-W.; Kim, Y.-H. Left atrial appendage: Morphology and function in patients with paroxysmal and persistent atrial fibrillation. Int. J. Cardiovasc. Imaging 2013, 29, 935–944. [Google Scholar] [CrossRef]
  36. Iwama, M.; Kawasaki, M.; Tanaka, R.; Ono, K.; Watanabe, T.; Hirose, T.; Nagaya, M.; Noda, T.; Watanabe, S.; Minatoguchi, S. Left atrial appendage emptying fraction assessed by a feature-tracking echocardiographic method is a determinant of thrombus in patients with nonvalvular atrial fibrillation. J. Cardiol. 2012, 59, 329–336. [Google Scholar] [CrossRef] [PubMed]
  37. Paliwal, N.; Park, H.-C.; Mao, Y.; Hong, S.J.; Lee, Y.; Spragg, D.D.; Calkins, H.; Trayanova, N.A. Slow blood-flow in the left atrial appendage is associated with stroke in atrial fibrillation patients. Heliyon 2024, 10, e26858. [Google Scholar] [CrossRef]
  38. Wang, L.; Wang, Z.; Fang, R.; Li, Z.-Y. Evaluation of Stroke Risk in Patients With Atrial Fibrillation Using Morphological and Hemodynamic Characteristics. Front. Cardiovasc. Med. 2022, 9, 842364. [Google Scholar] [CrossRef]
  39. García-Villalba, M.; Rossini, L.; Gonzalo, A.; Vigneault, D.; Martinez-Legazpi, P.; Durán, E.; Flores, O.; Bermejo, J.; McVeigh, E.; Kahn, A.M.; et al. Demonstration of patient-specific simulations to assess left atrial appendage thrombogenesis risk. Front. Physiol. 2021, 12, 596596. [Google Scholar] [CrossRef]
  40. Bosi, G.M.; Cook, A.; Rai, R.; Menezes, L.J.; Schievano, S.; Torii, R.; Burriesci, G. Computational fluid dynamic analysis of the left atrial appendage to predict thrombosis risk. Front. Cardiovasc. Med. 2018, 5, 34. [Google Scholar] [CrossRef]
  41. Corti, M.; Zingaro, A.; Dede’, L.; Quarteroni, A.M. Impact of atrial fibrillation on left atrium haemodynamics: A computational fluid dynamics study. Comput. Biol. Med. 2022, 150, 106143. [Google Scholar] [CrossRef]
  42. Lip, G.Y.; Nieuwlaat, R.; Pisters, R.; Lane, D.A.; Crijns, H.J. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: The euro heart survey on atrial fibrillation. Chest 2010, 137, 263–272. [Google Scholar] [CrossRef] [PubMed]
  43. Gage, B.; Waterman, A.; Shannon, W.; Boechler, M.; Rich, M.; Radford, M. Validation of clinical classification schemes for predicting stroke: Results from the national registry of atrial fibrillation. ACC Curr. J. Rev. 2001, 10, 2864–2870. [Google Scholar] [CrossRef]
  44. Hijazi, Z.; Lindbäck, J.; Alexander, J.H.; Hanna, M.; Held, C.; Hylek, E.M.; Lopes, R.D.; Oldgren, J.; Siegbahn, A.; Stewart, R.A.; et al. The ABC (age, biomarkers, clinical history) stroke risk score: A biomarker-based risk score for predicting stroke in atrial fibrillation. Eur. Heart J. 2016, 37, 1582–1590. [Google Scholar] [CrossRef] [PubMed]
  45. Wilson, P.W.; D’Agostino, R.B.; Levy, D.; Belanger, A.M.; Silbershatz, H.; Kannel, W.B. Prediction of coronary heart disease using risk factor categories. Circulation 1998, 97, 1837–1847. [Google Scholar] [CrossRef]
  46. Pisters, R.; Lane, D.A.; Nieuwlaat, R.; De Vos, C.B.; Crijns, H.J.; Lip, G.Y. A novel user-friendly score (has-bled) to assess 1-year risk of major bleeding in patients with atrial fibrillation: The euro heart survey. Chest 2010, 138, 1093–1100. [Google Scholar] [CrossRef]
  47. Fang, M.C.; Go, A.S.; Chang, Y.; Borowsky, L.; Pomernacki, N.K.; Singer, D.E. Comparison of risk stratification schemes to predict thromboembolism in people with nonvalvular atrial fibrillation. J. Am. Coll. Cardiol. 2008, 51, 810–815. [Google Scholar] [CrossRef]
  48. Hijazi, Z.; Oldgren, J.; Lindbäck, J.; Alexander, J.H.; Connolly, S.J.; Eikelboom, J.W.; Ezekowitz, M.D.; Held, C.; Hylek, E.M.; Lopes, R.D.; et al. The novel biomarker-based ABC (age, biomarkers, clinical history)-bleeding risk score for patients with atrial fibrillation: A derivation and validation study. Lancet 2016, 387, 2302–2311. [Google Scholar] [CrossRef]
  49. O’Brien, E.C.; Simon, D.N.; Thomas, L.E.; Hylek, E.M.; Gersh, B.J.; Ansell, J.E.; Kowey, P.R.; Mahaffey, K.W.; Chang, P.; Fonarow, G.C.; et al. The ORBIT bleeding score: A simple bedside score to assess bleeding risk in atrial fibrillation. Eur. Heart J. 2015, 36, 3258–3264. [Google Scholar] [CrossRef]
  50. Chen, J.-Y.; Zhang, A.-D.; Lu, H.-Y.; Guo, J.; Wang, F.-F.; Li, Z.-C. Chads2 versus cha2ds2-vasc score in assessing the stroke and thromboembolism risk stratification in patients with atrial fibrillation: A systematic review and meta-analysis. J. Geriatr. Cardiol. JGC 2013, 10, 258. [Google Scholar]
  51. Zhu, W.-G.; Xiong, Q.-M.; Hong, K. Meta-Analysis of CHADS2 versus CHA2DS2-VASc for Predicting Stroke and Thromboembolism in Atrial Fibrillation Patients Independent of Anticoagulation. Tex. Heart Inst. J. 2015, 42, 6–15. [Google Scholar] [CrossRef]
  52. Xiong, Q.; Chen, S.; Senoo, K.; Proietti, M.; Hong, K.; Lip, G.Y. The CHADS2 and CHA2DS2-VASc scores for predicting ischemic stroke among East Asian patients with atrial fibrillation: A systemic review and meta-analysis. Int. J. Cardiol. 2015, 195, 237–242. [Google Scholar] [CrossRef] [PubMed]
  53. Joundi, R.A.; Cipriano, L.E.; Sposato, L.A.; Saposnik, G.; on behalf of the Stroke Outcomes Research Working Group. Ischemic Stroke Risk in Patients With Atrial Fibrillation and CHA2DS2-VASc Score of 1. Stroke 2016, 47, 1364–1367. [Google Scholar] [CrossRef] [PubMed]
  54. Cai, Y.; Xiong, Q.; Chen, S.; Jiang, X.; Liao, J.; Chen, W.; Zou, L.; Su, L.; Zhu, Y.; Yin, Y.; et al. Left Atrial Appendage Thrombus in Patients with Nonvalvular Atrial Fibrillation before Catheter Ablation and Cardioversion: Risk Factors beyond the CHA2DS2-VASc Score. J. Cardiovasc. Dev. Dis. 2022, 9, 46. [Google Scholar] [CrossRef] [PubMed]
  55. Watson, T.; Shantsila, E.; Lip, G.Y. Mechanisms of thrombogenesis in atrial fibrillation: Virchow’s triad revisited. Lancet 2009, 373, 155–166. [Google Scholar] [CrossRef]
  56. Savelieva, I.; Bajpai, A.; Camm, A.J. Stroke in atrial fibrillation: Update on pathophysiology, new antithrombotic therapies, and evolution of procedures and devices. Ann. Med. 2007, 39, 371–391. [Google Scholar] [CrossRef]
  57. Wang, Y.; DI Biase, L.; Horton, R.P.; Nguyen, T.; Morhanty, P.; Natale, A. Left atrial appendage studied by computed tomography to help planning for appendage closure device placement. J. Cardiovasc. Electrophysiol. 2010, 21, 973–982. [Google Scholar] [CrossRef]
  58. García-Villarreal, A. Relationship between left atrial appendage morphology and thrombus formation in patients with atrial fibrillation and rheumatic mitral valve disease. Cirugía Card. En México 2024, 4, 79–83. [Google Scholar]
  59. Abanador-Kamper, N.; Bepperling, J.; Seyfarth, M.; Haage, P.; Kamper, L. Impact of left atrial appendage morphology on thrombus formation in TAVI patients with atrial fibrillation. Eur. J. Med. Res. 2023, 28, 88. [Google Scholar] [CrossRef]
  60. Kimura, T.; Takatsuki, S.; Inagawa, K.; Katsumata, Y.; Nishiyama, T.; Nishiyama, N.; Fukumoto, K.; Aizawa, Y.; Tanimoto, Y.; Tanimoto, K.; et al. Anatomical characteristics of the left atrial appendage in cardiogenic stroke with low CHADS2 scores. Heart Rhythm. 2013, 10, 921–925. [Google Scholar] [CrossRef]
  61. Korhonen, M.; Muuronen, A.; Arponen, O.; Mustonen, P.; Hedman, M.; Jäkälä, P.; Vanninen, R.; Taina, M. Left atrial appendage morphology in patients with suspected cardiogenic stroke without known atrial fibrillation. PLoS ONE 2015, 10, e0118822. [Google Scholar] [CrossRef]
  62. Khurram, I.M.; Dewire, J.; Mager, M.; Maqbool, F.; Zimmerman, S.L.; Zipunnikov, V.; Beinart, R.; Marine, J.E.; Spragg, D.D.; Berger, R.D.; et al. Relationship between left atrial appendage morphology and stroke in patients with atrial fibrillation. Heart Rhythm. 2013, 10, 1843–1849. [Google Scholar] [CrossRef] [PubMed]
  63. Yaghi, S.; Chang, A.D.; Akiki, R.; Collins, S.; Novack, T.; Hemendinger, M.; Schomer, A.; Mac Grory, B.; Cutting, S.; Burton, T.; et al. The left atrial appendage morphology is associated with embolic stroke subtypes using a simple classification system: A proof of concept study. J. Cardiovasc. Comput. Tomogr. 2020, 14, 27–33. [Google Scholar] [CrossRef]
  64. Słodowska, K.; Szczepanek, E.; Dudkiewicz, D.; Hołda, J.; Bolechała, F.; Strona, M.; Lis, M.; Batko, J.; Koziej, M.; Hołda, M.K. Morphology of the left atrial appendage: Introduction of a new simplified shape-based classification system. Heart Lung Circ. 2021, 30, 1014–1022. [Google Scholar] [CrossRef] [PubMed]
  65. Wongcharoen, W.; Tsao, H.; Wu, M.; Tai, C.; Chang, S.; Lin, Y.; Lo, L.; Chen, Y.; Sheu, M.; Chang, C.; et al. Morphologic Characteristics of the Left Atrial Appendage, Roof, and Septum: Implications for the Ablation of Atrial Fibrillation. J. Cardiovasc. Electrophysiol. 2006, 17, 951–956. [Google Scholar] [CrossRef] [PubMed]
  66. Yamamoto, M.; Seo, Y.; Kawamatsu, N.; Sato, K.; Sugano, A.; Machino-Ohtsuka, T.; Kawamura, R.; Nakajima, H.; Igarashi, M.; Sekiguchi, Y.; et al. Complex left atrial appendage morphology and left atrial appendage thrombus formation in patients with atrial fibrillation. Circ. Cardiovasc. Imaging 2014, 7, 337–343. [Google Scholar] [CrossRef]
  67. Adukauskaite, A.; Barbieri, F.; Senoner, T.; Plank, F.; Knoflach, M.; Boehme, C.; Hintringer, F.; Mueller, S.; Bauer, A.; Feuchtner, G.; et al. Left atrial appendage morphology and left atrial wall thickness are associated with cardio-embolic stroke. J. Clin. Med. 2020, 9, 3944. [Google Scholar] [CrossRef]
  68. Nedios, S.; Koutalas, E.; Kornej, J.; Rolf, S.; Arya, A.; Sommer, P.; Husser, D.; Hindricks, G.; Bollmann, A. Cardiogenic stroke despite low cha2ds2-vasc score: Assessing stroke risk by left atrial appendage anatomy (ask LAA). J. Cardiovasc. Electrophysiol. 2015, 26, 915–921. [Google Scholar] [CrossRef]
  69. Lei, C.; Gao, Q.; Wei, R.; Li, Q.; Liu, X.; Wu, L.; Yao, Y.; Fan, H.; Zheng, Z. Fractal geometry illustrated left atrial appendage morphology that predicted thrombosis and stroke in patients with atrial fibrillation. Front. Cardiovasc. Med. 2022, 9, 779528. [Google Scholar] [CrossRef]
  70. Yan, S.; Li, Q.; Xia, Z.; Yan, S.; Wei, Y.; Hong, K.; Wu, Y.; Li, J.; Cheng, X. Risk factors of thromboembolism in nonvalvular atrial fibrillation patients with low CHA2DS2-VASc score. Medicine 2019, 98, e14549. [Google Scholar] [CrossRef]
  71. Zhang, Y.; Yuan, Y.-Q. Value of left atrial diameter with cha2ds2-vasc score in predicting left atrial/left atrial appendage thrombosis in non-valvular atrial fibrillation. Arq. Bras. De Cardiol. 2021, 116, 325–331. [Google Scholar] [CrossRef] [PubMed]
  72. Zhou, M.; Chen, J.; Wang, H.; Xi, S.; Gan, T.; Zhao, L. Independent risk factors of atrial thrombosis in patients with nonvalvular atrial fibrillation and low cha 2 ds 2-vasc scores. Nan Fang Yi Ke Da Xue Xue Bao= J. South. Med. Univ. 2021, 41, 1243–1249. [Google Scholar]
  73. Chen, J.; Zhou, M.; Wang, H.; Zheng, Z.; Rong, W.; He, B.; Zhao, L. Risk factors for left atrial thrombus or spontaneous echo contrast in non-valvular atrial fibrillation patients with low CHA2DS2-VASc score. J. Thromb. Thrombolysis 2022, 53, 523–531. [Google Scholar] [CrossRef] [PubMed]
  74. Qureshi, A.; Lip, G.Y.H.; Nordsletten, D.A.; Williams, S.E.; Aslanidi, O.; de Vecchi, A. Imaging and biophysical modelling of thrombogenic mechanisms in atrial fibrillation and stroke. Front. Cardiovasc. Med. 2023, 9, 1074562. [Google Scholar] [CrossRef]
  75. Furie, B.; Furie, B.C. Mechanisms of thrombus formation. N. Engl. J. Med. 2008, 359, 938–949. [Google Scholar] [CrossRef]
  76. Hoffman, M. Remodeling the blood coagulation cascade. J. Thromb. Thrombolysis 2003, 16, 17–20. [Google Scholar] [CrossRef]
  77. Sun, Y.; Ling, Y.; Chen, Z.; Wang, Z.; Li, T.; Tong, Q.; Qian, Y. Finding low CHA2DS2-VASc scores unreliable? Why not give morphological and hemodynamic methods a try? Front. Cardiovasc. Med. 2023, 9, 1032736. [Google Scholar] [CrossRef]
  78. Demirkiran, A.; Amier, R.P.; Hofman, M.B.M.; van der Geest, R.J.; Robbers, L.F.H.J.; Hopman, L.H.G.A.; Mulder, M.J.; van de Ven, P.; Allaart, C.P.; van Rossum, A.C.; et al. Altered left atrial 4D flow characteristics in patients with paroxysmal atrial fibrillation in the absence of apparent remodeling. Sci. Rep. 2021, 11, 1–11. [Google Scholar] [CrossRef]
  79. Ferdian, E.; Dubowitz, D.J.; Mauger, C.A.; Wang, A.; Young, A.A. WSSNet: Aortic Wall Shear Stress Estimation Using Deep Learning on 4D Flow MRI. Front. Cardiovasc. Med. 2022, 8, 769927. [Google Scholar] [CrossRef]
  80. Lee, Y.; Park, H.-C.; Lee, Y.; Kim, S.-G. Comparison of morphologic features and flow velocity of the left atrial appendage among patients with atrial fibrillation alone, transient ischemic attack, and cardioembolic stroke. Am. J. Cardiol. 2017, 119, 1596–1604. [Google Scholar] [CrossRef]
  81. Fukushima, N.; Kato, K.; Ejima, K.; Sato, H.; Fukushima, K.; Saito, C.; Hayashi, K.; Arai, K.; Manaka, T.; Ashihara, K.; et al. Correlation between left atrial appendage morphology and flow velocity in patients with paroxysmal atrial fibrillation. Eur. Heart J.-Cardiovasc. Imaging 2016, 17, 59–66. [Google Scholar] [CrossRef]
  82. Turgut, N.; Akdemir, O.; Turgut, B.; Demir, M.; Ekuklu, G.; Vural, Ö.; Özbay, G.; Utku, U. Hypercoagulopathy in stroke patients with nonvalvular atrial fibrillation: Hematologic and cardiologic investigations. Clin. Appl. Thromb. 2006, 12, 15–20. [Google Scholar] [CrossRef]
  83. Akar, J.G.; Jeske, W.; Wilber, D.J. Acute onset human atrial fibrillation is associated with local cardiac platelet activation and endothelial dysfunction. J. Am. Coll. Cardiol. 2008, 51, 1790–1793. [Google Scholar] [CrossRef]
  84. Marin, F.; Roldán, V.; Climent, V.E.; Ibáñez, A.; García, A.; Marco, P.; Sogorb, F.; Lip, G.Y.H. Plasma von willebrand factor, soluble thrombomodulin, and fibrin D-dimer concentrations in acute onset non-rheumatic atrial fibrillation. Heart 2004, 90, 1162–1166. [Google Scholar] [CrossRef] [PubMed]
  85. Yau, J.W.; Teoh, H.; Verma, S. Endothelial cell control of thrombosis. BMC Cardiovasc. Disord. 2015, 15, 130. [Google Scholar] [CrossRef] [PubMed]
  86. Goldberg, R.J.; Nakagawa, T.; Johnson, R.J.; Thurman, J.M. The role of endothelial cell injury in thrombotic microangiopathy. Am. J. Kidney Dis. 2010, 56, 1168–1174. [Google Scholar] [CrossRef] [PubMed]
  87. Park, J.; Joung, B.; Uhm, J.-S.; Shim, C.Y.; Hwang, C.; Lee, M.H.; Pak, H.-N. High left atrial pressures are associated with advanced electroanatomical remodeling of left atrium and independent predictors for clinical recurrence of atrial fibrillation after catheter ablation. Heart Rhythm. 2014, 11, 953–960. [Google Scholar] [CrossRef]
  88. Linhart, M.; Lewalter, T.; Mittmann-Braun, E.L.; Karbach, N.C.; Andrié, R.P.; Hammerstingl, C.; Fimmers, R.; Kreuz, J.; Nickenig, G.; Schrickel, J.W.; et al. Left atrial pressure as predictor for recurrence of atrial fibrillation after pulmonary vein isolation. J. Interv. Card. Electrophysiol. 2013, 38, 107–114. [Google Scholar] [CrossRef]
  89. Manning, W.J.; Weintraub, R.M.; Waksmonski, C.A.; Haering, J.M.; Rooney, P.S.; Maslow, A.D.; Johnson, R.G.; Douglas, P.S. Accuracy of transesophageal echocardiography for identifying left atrial thrombi: A prospective, intraoperative study. Ann. Intern. Med. 1995, 123, 817–822. [Google Scholar] [CrossRef]
  90. Mojadidi, M.K.; Bogush, N.; Caceres, J.D.; Msaouel, P.; Tobis, J.M. Diagnostic accuracy of transesophageal echocardiogram for the detection of patent foramen ovale: A meta-analysis. Echocardiography 2014, 31, 752–758. [Google Scholar] [CrossRef]
  91. Abdelmoneim, S.S.; Mulvagh, S.L. Techniques to improve left atrial appendage imaging. J. Atr. Fibrillation 2014, 7, 1059. [Google Scholar] [CrossRef]
  92. Peeters, Y.; Bernards, J.; Mekeirele, M.; Hoffmann, B.; Raes, M.; Malbrain, M. Hemodynamic monitoring: To calibrate or not to calibrate? part 1–calibrated techniques. Anestezjol. Intensywna Ter. 2019, 47, 503–516. [Google Scholar] [CrossRef] [PubMed]
  93. Sun, A.; Ren, S.; Xiao, Y.; Chen, Y.; Wang, N.; Li, C.; Tan, X.; Pan, Y.; Sun, F.; Ren, W. Real-time 3D echocardiographic transilluminated imaging combined with artificially intelligent left atrial appendage measurement for atrial fibrillation interventional procedures. Front. Physiol. 2022, 13, 1043551. [Google Scholar] [CrossRef] [PubMed]
  94. Farkowski, M.M.; Jubele, K.; Marín, F.; Gandjbakhch, E.; Ptaszynski, P.; Merino, J.L.; Lenarczyk, R.; Potpara, T.S. Diagnosis and management of left atrial appendage thrombus in patients with atrial fibrillation undergoing cardioversion or percutaneous left atrial procedures: Results of the European heart rhythm association survey. EP Eur. 2020, 22, 162–169. [Google Scholar] [CrossRef] [PubMed]
  95. Sobrino, A.; Tzikas, A.; Freixa, X.; Pulido, A.; Chan, J.; Garceau, P.; Ibrahim, R.; Basmadjian, A.J. Intra-procedural imaging of the left atrial appendage: Implications for closure with the Amplatzer™ cardiac plug. Arch. Cardiol. Mex. 2014, 84, 17–24. [Google Scholar] [CrossRef]
  96. Brown, L.; Brown, A. Transesophageal echocardiography: Implications for the critical care nurse. Crit. Care Nurse 1994, 14, 55–59. [Google Scholar] [CrossRef]
  97. Hur, J.; Kim, Y.J.; Nam, J.E.; Choe, K.O.; Choi, E.Y.; Shim, C.Y.; Choi, B.W. Thrombus in the left atrial appendage in stroke patients: Detection with cardiac CT angiography—A preliminary report. Radiology 2008, 249, 81–87. [Google Scholar] [CrossRef]
  98. Romero, J.; Husain, S.A.; Kelesidis, I.; Sanz, J.; Medina, H.M.; Garcia, M.J. Detection of left atrial appendage thrombus by cardiac computed tomography in patients with atrial fibrillation: A meta-analysis. Circ. Cardiovasc. Imaging 2013, 6, 185–194. [Google Scholar] [CrossRef]
  99. Nakajima, H.; Seo, Y.; Ishizu, T.; Yamamoto, M.; Machino, T.; Harimura, Y.; Kawamura, R.; Sekiguchi, Y.; Tada, H.; Aonuma, K. Analysis of the left atrial appendage by three-dimensional transesophageal echocardiography. Am. J. Cardiol. 2010, 106, 885–892. [Google Scholar] [CrossRef]
  100. Schutt, E.G.; Klein, D.H.; Mattrey, R.M.; Riess, J.G. Injectable microbubbles as contrast agents for diagnostic ultrasound imaging: The key role of perfluorochemicals. Angew. Chem. Int. Ed. Engl. 2003, 42, 3218–3235. [Google Scholar] [CrossRef]
  101. Sun, L.; Li, Y.; Zhang, Y.T.; Shen, J.X.; Xue, F.H.; Cheng, H.D.; Qu, X.F. A computer-aided diagnostic algorithm improves the accuracy of transesophageal echocardiography for left atrial thrombi: A single-center prospective study. J. Ultrasound Med. 2014, 33, 83–91. [Google Scholar] [CrossRef] [PubMed]
  102. Akella, K.; Murtaza, G.; Turagam, M.; Sharma, S.; Madoukh, B.; Amin, A.; Gopinathannair, R.; Lakkireddy, D. Evaluating the role of transesophageal echocardiography (TEE) or intracardiac echocardiography (ICE) in left atrial appendage occlusion: A meta-analysis. J. Interv. Card. Electrophysiol. 2021, 60, 41–48. [Google Scholar] [CrossRef] [PubMed]
  103. Desimone, C.V.; Asirvatham, S.J. Ice imaging of the left atrial appendage. J. Cardiovasc. Electrophysiol. 2014, 25, 1272. [Google Scholar] [CrossRef]
  104. Baran, J.; Stec, S.; Pilichowska-Paszkiet, E.; Zaborska, B.; Sikora-Frac, M.; Kryski, T.; Michaowska, I.; Łopatka, R.; Kuakowski, P. Intracardiac echocardiography for detection of thrombus in the left atrial appendage: Comparison with transesophageal echocardiography in patients undergoing ablation for atrial fibrillation: The action-ice i study. Circ. Arrhythmia Electrophysiol. 2013, 6, 1074–1081. [Google Scholar] [CrossRef]
  105. Mojadidi, M.K.; Winoker, J.S.; Roberts, S.C.; Msaouel, P.; Zaman, M.O.; Gevorgyan, R.; Tobis, J.M. Accuracy of conventional transthoracic echocardiography for the diagnosis of intracardiac right-to-left shunt: A meta-analysis of prospective studies. Echocardiography 2014, 31, 1036–1048. [Google Scholar] [CrossRef]
  106. Bai, A.D.; Steinberg, M.; Showler, A.; Burry, L.; Bhatia, R.S.; Tomlinson, G.A.; Bell, C.M.; Morris, A.M. Diagnostic accuracy of transthoracic echocardiography for infective endocarditis findings using transesophageal echocardiography as the reference standard: A meta-analysis. J. Am. Soc. Echocardiogr. 2017, 30, 639–646.e8. [Google Scholar] [CrossRef] [PubMed]
  107. Kumar, V.; Nanda, N.C. Is it time to move on from two-dimensional transesophageal to three-dimensional transthoracic echocardiography for assessment of left atrial appendage? review of existing literature. Echocardiography 2012, 29, 112–116. [Google Scholar] [CrossRef]
  108. Karakus, G.; Kodali, V.; Inamdar, V.; Nanda, N.C.; Suwanjutah, T.; Pothineni, K.R. Comparative assessment of left atrial appendage by transesophageal and combined two-and three-dimensional transthoracic echocardiography. Echocardiography 2008, 25, 918–924. [Google Scholar] [CrossRef]
  109. Busse, A.; Cantré, D.; Beller, E.; Streckenbach, F.; Öner, A.; Ince, H.; Weber, M.-A.; Meinel, F.G. Cardiac CT: Why, when, and how: Update 2019. Der Radiol. 2019, 59 (Suppl. 1), 1–9. [Google Scholar] [CrossRef]
  110. Hamada, M. Left ventricular thrombus in hypertrophic cardiomyopathy. Intern. Med. 2019, 58, 465–467. [Google Scholar] [CrossRef]
  111. Pontone, G.; Andreini, D.; Bartorelli, A.L.; Cortinovis, S.; Mushtaq, S.; Bertella, E.; Annoni, A.; Formenti, A.; Nobili, E.; Trabattoni, D.; et al. Diagnostic accuracy of coronary computed tomography angiography: A comparison between prospective and retrospective electrocardiogram triggering. J. Am. Coll. Cardiol. 2009, 54, 346–355. [Google Scholar] [CrossRef] [PubMed]
  112. Corti, R.; Osende, J.I.; Fayad, Z.A.; Fallon, J.T.; Fuster, V.; Mizsei, G.; Dickstein, E.; Drayer, B.; Badimon, J.J. In vivo noninvasive detection and age definition of arterial thrombus by MRI. J. Am. Coll. Cardiol. 2002, 39, 1366–1373. [Google Scholar] [CrossRef]
  113. Burrell, L.D.; Horne, B.D.; Anderson, J.L.; Muhlestein, J.B.; Whisenant, B.K. Usefulness of left atrial appendage volume as a predictor of embolic stroke in patients with atrial fibrillation. Am. J. Cardiol. 2013, 112, 1148–1152. [Google Scholar] [CrossRef] [PubMed]
  114. Rajiah, P.S.; François, C.J.; Leiner, T. Cardiac MRI: State of the art. Radiology 2023, 307, e223008. [Google Scholar] [CrossRef]
  115. Chang, P.; Xiao, J.; Hu, Z.; Kwan, A.C.; Fan, Z. Imaging of left heart intracardiac thrombus: Clinical needs, current imaging, and emerging cardiac magnetic resonance techniques. Ther. Adv. Cardiovasc. Dis. 2022, 16, 17539447221107737. [Google Scholar] [CrossRef] [PubMed]
  116. Mohrs, O.K.; Ruebesam, D.; Peters, J. Computed tomography in a patient after percutaneous left atrial appendage transcatheter occlusion (PLAATO). Heart 2006, 92, 486. [Google Scholar] [CrossRef]
  117. Soulat, G.; McCarthy, P.; Markl, M. 4D flow with MRI. Annu. Rev. Biomed. Eng. 2020, 22, 103–126. [Google Scholar] [CrossRef]
  118. Cherry, M.; Khatir, Z.; Khan, A.; Bissell, M. The impact of 4D-Flow MRI spatial resolution on patient-specific CFD simulations of the thoracic aorta. Sci. Rep. 2022, 12, 15128. [Google Scholar] [CrossRef]
  119. Markl, M.; Frydrychowicz, A.; Kozerke, S.; Hope, M.; Wieben, O. 4D flow MRI. J. Magn. Reson. Imaging 2012, 36, 1015–1036. [Google Scholar] [CrossRef]
  120. Markl, M.; Lee, D.C.; Ng, J.; Carr, M.; Carr, J.; Goldberger, J.J. Left atrial 4-dimensional flow magnetic resonance imaging: Stasis and velocity mapping in patients with atrial fibrillation. Investig. Radiol. 2016, 51, 147–154. [Google Scholar] [CrossRef]
  121. Callaghan, F.M.; Grieve, S.M. Spatial resolution and velocity field improvement of 4D-flow MRI. Magn. Reson. Med. 2017, 78, 1959–1968. [Google Scholar] [CrossRef] [PubMed]
  122. Wåhlin, A.; Ambarki, K.; Birgander, R.; Wieben, O.; Johnson, K.; Malm, J.; Eklund, A. Measuring pulsatile flow in cerebral arteries using 4D phase-contrast MR imaging. Am. J. Neuroradiol. 2013, 34, 1740–1745. [Google Scholar] [CrossRef]
  123. Morales, X.; Mill, J.; Juhl, K.A.; Olivares, A.; Jimenez-Perez, G.; Paulsen, R.R.; Camara, O.T. Deep learning surrogate of computational fluid dynamics for thrombus formation risk in the left atrial appendage. In Statistical Atlases and Computational Models of the Heart. Multi-Sequence CMR Segmentation, CRT-EPiggy and LV Full Quantification Challenges; Pop, M., Sermesant, M., Camara, O., Zhuang, X., Li, S., Young, A., Mansi, T., Suinesiaputra, A., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 157–166. [Google Scholar]
  124. Gasparotti, E.; Fanni, B.M.; Del Pia, E.; Capellini, K.; Danielli, F.; Berti, F.; Clemente, A.; Berti, S.; Pennnati, G.; Petrini, L.; et al. Computational fluid dynamic simulation to evaluate the device-related effects after left atrial appendage occlusion. In Proceedings of the International Symposium on Computer Methods in Biomechanics and Biomedical Engineering, Paris, France, 3–5 May 2023; Springer: Berlin/Heidelberg, Germany, 2023; pp. 205–212. [Google Scholar]
  125. Masci, A.; Dalessandro, N.; Scivoletto, A.; Severi, S.; Ansaloni, F.; Monte, A.D.; Tomasi, C.; Fabbri, C.; Corsi, C. Endocardial left atrial appendage occlusion in atrial fibrillation: Computational fluid dynamics simulations to assess stroke risk. EP Eur. 2021, 23 (Suppl. S3), euab116–euab173. [Google Scholar] [CrossRef]
  126. Ghodrati-Misek, M.; Schlöglhofer, T.; Gross, C.; Maurer, A.; Zimpfer, D.; Beitzke, D.; Zonta, F.; Moscato, F.; Schima, H.; Aigner, P. Left atrial appendage occlusion in ventricular assist device patients to decrease thromboembolic events: A computer simulation study. Front. Physiol. 2022, 13, 1010862. [Google Scholar] [CrossRef]
  127. Mill, J.; Agudelo, V.; Olivares, A.L.; Pons, M.I.; Silva, E.; Nuñez-Garcia, M.; Morales, X.; Arzamendi, D.; Freixa, X.; Noailly, J.; et al. Sensitivity analysis of in silico fluid simulations to predict thrombus formation after left atrial appendage occlusion. Mathematics 2021, 9, 2304. [Google Scholar] [CrossRef]
  128. Bäck, S.; Lantz, J.; Skoda, I.; Henriksson, L.; Persson, A.; Karlsson, L.O.; Carlhäll, C.-J.; Ebbers, T. Comprehensive left atrial flow component analysis reveals abnormal flow patterns in paroxysmal atrial fibrillation. Am. J. Physiol. Circ. Physiol. 2024, 326, H511–H521. [Google Scholar] [CrossRef]
  129. Dillon-Murphy, D.; Marlevi, D.; Ruijsink, B.; Qureshi, A.; Chubb, H.; Kerfoot, E.; O’Neill, M.; Nordsletten, D.; Aslanidi, O.; de Vecchi, A. Modeling left atrial flow, energy, blood heating distribution in response to catheter ablation therapy. Front. Physiol. 2018, 9, 1757. [Google Scholar] [CrossRef]
  130. Kelly, P.J.; Albers, G.W.; Chatzikonstantinou, A.; De Marchis, G.M.; Ferrari, J.; George, P.; Katan, M.; Knoflach, M.; Kim, J.S.; Li, L.; et al. Validation and comparison of imaging-based scores for prediction of early stroke risk after transient ischaemic attack: A pooled analysis of individual-patient data from cohort studies. Lancet Neurol. 2016, 15, 1238–1247. [Google Scholar] [CrossRef]
  131. Otani, T.; Al-Issa, A.; Pourmorteza, A.; McVeigh, E.R.; Wada, S.; Ashikaga, H. A computational framework for personalized blood flow analysis in the human left atrium. Ann. Biomed. Eng. 2016, 44, 3284–3294. [Google Scholar] [CrossRef]
  132. Brady, P.; Kelion, A.; Hyde, T.; Barnes, E.; Rahbi, H.; Beale, A.; Ramcharitar, S. CT coronary angiography with HeartFlow®: A user’s perspective. Br. J. Cardiol. 2019, 26, 105–109. [Google Scholar]
  133. Cai, S.; Lu, Y.; Li, B.; Gao, Q.; Xu, L.; Hu, X.; Zhang, L. Segmentation of cardiac tissues and organs for CCTA images based on a deep learning model. Front. Phys. 2023, 11, 1266500. [Google Scholar] [CrossRef]
  134. Zheng, X.; Wang, F.; Zhang, J.; Cui, X.; Jiang, F.; Chen, N.; Zhou, J.; Chen, J.; Lin, S.; Zou, J. Using machine learning to predict atrial fibrillation diagnosed after ischemic stroke. Int. J. Cardiol. 2022, 347, 21–27. [Google Scholar] [CrossRef]
  135. Chen, L.; Huang, S.-H.; Wang, T.-H.; Lan, T.-Y.; Tseng, V.S.; Tsao, H.-M.; Wang, H.-H.; Tang, G.-J. Deep learning-based automatic left atrial appendage filling defects assessment on cardiac computed tomography for clinical and subclinical atrial fibrillation patients. Heliyon 2023, 9, e12945. [Google Scholar] [CrossRef]
  136. Gómez, C.; Daniel; Ferez, X.M.; Rey, O.C. Geometrical Deep Learning for the Estimation of Residence Time in the Left Atria. In International Workshop on Statistical Atlases and Computational Models of the Heart; Springer Nature Switzerland: Cham, Switzerland, 2022. [Google Scholar]
  137. Ryu, W.-S.; Schellingerhout, D.; Lee, H.; Lee, K.-J.; Kim, C.K.; Kim, B.J.; Chung, J.-W.; Lim, J.-S.; Kim, J.-T.; Kim, D.-H.; et al. Deep learning-based automatic classification of ischemic stroke subtype using diffusion-weighted images. J. Stroke 2024, 26, 300–311. [Google Scholar] [CrossRef] [PubMed]
  138. Ronneberger, O.; Fischer, P.; Brox, T. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, 5–9 October 2015; Proceedings, Part III 18; Springer: Berlin/Heidelberg, Germany, 2015; pp. 234–241. [Google Scholar]
  139. Siddique, N.; Paheding, S.; Elkin, C.P.; Devabhaktuni, V. U-net and its variants for medical image segmentation: A review of theory and applications. IEEE Access 2021, 9, 82031–82057. [Google Scholar] [CrossRef]
  140. Zhou, H.; He, L.; Zhang, Y.; Shen, L.; Chen, B. Interpretable graph convolutional network of multi-modality brain imaging for alzheimer’s disease diagnosis. In Proceedings of the 2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI), IEEE, Kolkata, India, 28–31 March 2022; pp. 1–5. [Google Scholar]
  141. Xia, J.; Chan, Y.H.; Girish, D.; Rajapakse, J.C. IMG-GCN: Interpretable modularity-guided structure-function interactions learning for brain cognition and disorder analysis. In Proceedings of the International Conference on Medical Image Computing and Computer-Assisted Intervention, Marrakesh, Morocco, 7–11 October 2024; Springer: Berlin/Heidelberg, Germany, 2024; pp. 470–480. [Google Scholar]
  142. Xia, J.; Chan, Y.H.; Girish, D.; Rajapakse, J.C.; Han, W.; Kang, X.; He, W.; Jiang, L.; Li, H.; Xu, B. A new method for disease diagnosis based on hierarchical BRB with power set. Heliyon 2023, 9, e13619. [Google Scholar] [CrossRef]
  143. Yang, Y.; Zhang, Z.; Ng, C.Y.; Li, G.; Liu, T. Meta-analysis of CHADS2 Score in Predicting Atrial Fibrillation. Am. J. Cardiol. 2015, 116, 554–562. [Google Scholar] [CrossRef]
  144. Zeng, J.; Yu, P.; Cui, W.; Wang, X.; Ma, J.; Zeng, C. Comparison of HAS-BLED with other risk models for predicting the bleeding risk in anticoagulated patients with atrial fibrillation A PRISMA-compliant article. Medicine 2020, 99, e20782. [Google Scholar] [CrossRef]
  145. Chang, G.; Xie, Q.; Ma, L.; Hu, K.; Zhang, Z.; Mu, G.; Cui, Y. Accuracy of HAS-BLED and other bleeding risk assessment tools in predicting major bleeding events in atrial fibrillation: A network meta-analysis. J. Thromb. Haemost. 2020, 18, 791–801. [Google Scholar] [CrossRef]
  146. Sun, P.; Guo, Z.H.; Zhang, H.B. CHA2DS2-VASc Score as a Predictor for Left Atrial Thrombus or Spontaneous Echo Contrast in Patients with Nonvalvular Atrial Fibrillation: A Meta-Analysis. BioMed Res. Int. 2020, 2020, 2679539. [Google Scholar] [CrossRef]
Applsci 15 04633 g001
Figure 1. Components of Virchow’s triad for thrombogenesis in atrial fibrillation [55]. The upward arrow indicates an increase in the content of the corresponding substance.
Figure 1. Components of Virchow’s triad for thrombogenesis in atrial fibrillation [55]. The upward arrow indicates an increase in the content of the corresponding substance.
Applsci 15 04633 g001
Applsci 15 04633 g002
Figure 2. TEE and CT images of LAA and LA: (A). the LAA is assessed in 90° TEE views [95]; (B). the LAA is assessed in 120° TEE views [95]; (C). TEE operation demonstration [96]; (D). CT images of LA and LAA, with dashed lines representing the long axis of LAA [97].
Figure 2. TEE and CT images of LAA and LA: (A). the LAA is assessed in 90° TEE views [95]; (B). the LAA is assessed in 120° TEE views [95]; (C). TEE operation demonstration [96]; (D). CT images of LA and LAA, with dashed lines representing the long axis of LAA [97].
Applsci 15 04633 g002
Applsci 15 04633 g003
Figure 3. The CFD results of LA/LAA (streamlines of blood flow during a representative cardiac cycle in the posterior view) for two patients (case 1 and case 2) with AF history [131].
Figure 3. The CFD results of LA/LAA (streamlines of blood flow during a representative cardiac cycle in the posterior view) for two patients (case 1 and case 2) with AF history [131].
Applsci 15 04633 g003
Applsci 15 04633 g004
Figure 4. Pipeline of the CFD method (blue arrows) and AI method (orange arrows) for LA/LAA related tasks. a. non-AI methods for model reconstruction and segmentation from medical imaging; b. CFD methods; c. and d. expert judgement. A. AI methods for model reconstruction and segmentation from medical imaging; B. AI technology alternative to CFD; C. AI technologies that directly diagnose stroke subtypes or assess stroke risk scores based on medical imaging.
Figure 4. Pipeline of the CFD method (blue arrows) and AI method (orange arrows) for LA/LAA related tasks. a. non-AI methods for model reconstruction and segmentation from medical imaging; b. CFD methods; c. and d. expert judgement. A. AI methods for model reconstruction and segmentation from medical imaging; B. AI technology alternative to CFD; C. AI technologies that directly diagnose stroke subtypes or assess stroke risk scores based on medical imaging.
Applsci 15 04633 g004
Table 1. Description and performance of several stroke risk stratifications.
Table 1. Description and performance of several stroke risk stratifications.
Stroke Risk StratificationCHADS2FraminghamCHA2DS2-VAScABC
Risk factorsPatient CharacteristicsAge✓ ✓✓
Sex ✓
ComorbiditiesDiabetes✓✓✓
Hypertension✓ ✓
Vascular Diseases ✓✓
Heart Failure✓ ✓
Prior Stroke/TIA✓✓✓✓
Prior Thromboembolism ✓
BiomarkersNT-ProBNP ✓
Hs-CTnI ✓
C-statistic range [mean value]0.61–0.67
[0.645]		0.633
[0.633]		0.61–0.71
[0.644]		0.6–0.8
[0.642]
References[22,23,24,25][25][11,22,23,24,25,26,27,28,29][11,23,25]
Table 2. Description and performance of several bleeding risk stratifications.
Table 2. Description and performance of several bleeding risk stratifications.
Stroke Risk StratificationHAS-BLEDATRIAORBITABS-Bleeding
Risk factorsPatient
Characteristics		Age✓✓✓✓
ComorbiditiesDrug/Alcohol Use✓
Hypertension✓✓
Abnormal renal of liver function✓✓✓
Anemia ✓✓
Previous Stroke✓
Labile INR✓
Bleeding history or predisposition✓✓✓✓
BiomarkersHemoglobin ✓
cTn-hs ✓
GDF-15 or cystatin C/CKD-EPI ✓
C-statistic range [mean value]0.62–0.63
[0.628]		0.63–0.683
[0.651]		0.61–0.65
[0.635]		0.66–0.68
[0.67]
References[22,23,30,31,32][22,23,25,27][22,23,25,27][11,32]
Table 3. LAA hemodynamic indexes related to the risk of stroke and thrombosis. Ved is left atrial appendage end-diastolic volume and Ves is left atrial appendage end-systolic volume. T is the cardiac cycle. τw is the WSS. u, v, and w are the three components of velocity. TAWSS is time-averaged WSS. OSI is the oscillatory shear index. ECAP is endothelial cell activation potential. RRT is relative residence time. SSR is the shear strain rate. FS is the flow stasis index, representing the fraction of time of the heartbeat in which the velocity magnitude in a specific point is smaller than 0.1 m/s. m1 is the mean age field of red blood cells. P and N denote the positive and negative correlations, alternatively.
Table 3. LAA hemodynamic indexes related to the risk of stroke and thrombosis. Ved is left atrial appendage end-diastolic volume and Ves is left atrial appendage end-systolic volume. T is the cardiac cycle. τw is the WSS. u, v, and w are the three components of velocity. TAWSS is time-averaged WSS. OSI is the oscillatory shear index. ECAP is endothelial cell activation potential. RRT is relative residence time. SSR is the shear strain rate. FS is the flow stasis index, representing the fraction of time of the heartbeat in which the velocity magnitude in a specific point is smaller than 0.1 m/s. m1 is the mean age field of red blood cells. P and N denote the positive and negative correlations, alternatively.
LAA Hemodynamic IndexDefinitionRelationReference
Velocity-N[15,33,34]
Vortex size-P[7]
Vorticity-N[34]
Ejection fraction V e d − V e s V e d N[35,36]
TAWSS 1 T ∫ 0 T τ w d t N[19,20,33,37]
OSI 1 2 1 − ∫ 0 T τ w d t ∫ 0 T τ w d t P[19,20,33]
ECAP OSI ECAP P[19,20,37]
RRT [ 1 − 2 × O S I × T A W S S ] − 1 P[19,20,38]
Kinetic energy 1 2 u 2 + v 2 + w 2 N[34,39]
SSR-N[40]
AS F S x m 1 x T P[41]
M4the fourth moment of the blood age probability distributionP[14]
Table 4. Research on left atrial appendage based on artificial intelligence methods. CCT: cardiac computed tomography. CCTA: cardiac computed tomography angiography. ICC: intraclass correlation coefficient. DSC: dice similarity coefficient. NRMSE: normalized root mean square error. MAE: mean absolute error. NMAE: normalized mean absolute error. DNN: deep neural network.
Table 4. Research on left atrial appendage based on artificial intelligence methods. CCT: cardiac computed tomography. CCTA: cardiac computed tomography angiography. ICC: intraclass correlation coefficient. DSC: dice similarity coefficient. NRMSE: normalized root mean square error. MAE: mean absolute error. NMAE: normalized mean absolute error. DNN: deep neural network.
AI TaskReferenceNeural NetworkInputOutputPerformance
A[135]U-netCCTFelling defects areaAUC = 0.979
ICC ≥ 0.895
A[133]3D-UnetCCTALV, Aorta, LA,
LAA and Myoc		DSC = 0.882–0.961
B[21]PCA-FCN/U-net/GDL2D mapping of the 3D geometrical model for PCA-FCN U-net and point clouds of 3D geometricalECAPGDL has the best performance MAE = 0.5
B[136]GDL/U-net3D point clouds for GDL and
mesh information for U-net		RRTNRMSE = 0.08
for both two NNs
B[20]GDL3D point cloudsTAWSS/OSI/ECAP/RRTMAE = 0.784
B[19]GDL3D point cloudsVelocity and
Pressure		NMAE = 5.7%
for pressure and 10.35% for velocity
C[134]DNNMedical recordrisk score for
post-stroke AF		AUC = 0.922
C[137]U-net
EfficientNetV2		MRI with AF informationStroke subtypepercentage agreement with expert consensus = 72.9%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Liu, X.; Gao, Q. Study of Atrial Fibrillation and Stroke Based on Geometrical and Hemodynamic Characteristics: A Review. Appl. Sci. 2025, 15, 4633. https://doi.org/10.3390/app15094633

AMA Style

Liu X, Gao Q. Study of Atrial Fibrillation and Stroke Based on Geometrical and Hemodynamic Characteristics: A Review. Applied Sciences. 2025; 15(9):4633. https://doi.org/10.3390/app15094633

Chicago/Turabian Style

Liu, Xiaoyu, and Qi Gao. 2025. "Study of Atrial Fibrillation and Stroke Based on Geometrical and Hemodynamic Characteristics: A Review" Applied Sciences 15, no. 9: 4633. https://doi.org/10.3390/app15094633

APA Style

Liu, X., & Gao, Q. (2025). Study of Atrial Fibrillation and Stroke Based on Geometrical and Hemodynamic Characteristics: A Review. Applied Sciences, 15(9), 4633. https://doi.org/10.3390/app15094633

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop