Left atrial appendage segmentation from 3D CCTA images for occluder placement procedure

Accurate fully automated assessment of left ventricle, left atrium, and left atrial appendage function from computed tomography using deep learning

Aims

Assessment of cardiac function is essential for diagnosis and treatment planning in cardiovascular disease. Volume of cardiac regions and the derived measures of stroke volume (SV) and ejection fraction (EF) are most accurately calculated from imaging. This study aims to develop a fully automatic deep learning approach for calculation of cardiac function from computed tomography (CT).

Methods and results

Time-resolved CT data sets from 39 patients were used to train segmentation models for the left side of the heart including the left ventricle (LV), left atrium (LA), and left atrial appendage (LAA). We compared nnU-Net, 3D TransUNet, and UNETR. Dice Similarity Scores (DSS) were similar between nnU-Net (average DSS = 0.91) and 3D TransUNet (DSS = 0.89) while UNETR performed less well (DSS = 0.69). Intra-class correlation analysis showed nnU-Net and 3D TransUNet both accurately estimated LVSV (ICCnnU-Net = 0.95; ICC3DTransUNet = 0.94), LVEF (ICCnnU-Net = 1.00; ICC3DTransUNet = 1.00), LASV (ICCnnU-Net = 0.91; ICC3DTransUNet = 0.80), LAEF (ICCnnU-Net = 0.95; ICC3DTransUNet = 0.81), and LAASV (ICCnnU-Net = 0.79; ICC3DTransUNet = 0.81). Only nnU-Net significantly predicted LAAEF (ICCnnU-Net = 0.68). UNETR was not able to accurately estimate cardiac function. Time to convergence during training and time needed for inference were both faster for 3D TransUNet than for nnU-Net.

Conclusion

nnU-Net outperformed two different vision transformer architectures for the segmentation and calculation of function parameters for the LV, LA, and LAA. Fully automatic calculation of cardiac function parameters from CT using deep learning is fast and reliable.

Pulmonary vein flow split effects in patient-specific simulations of left atrial flow

Disruptions to left atrial (LA) blood flow, such as those caused by atrial fibrillation (AF), can lead to thrombosis in the left atrial appendage (LAA) and an increased risk of systemic embolism. LA hemodynamics are influenced by various factors, including LA anatomy and function, and pulmonary vein (PV) inflow conditions. In particular, the PV flow split can vary significantly among and within patients depending on multiple factors. In this study, we investigated how changes in PV flow split affect LA flow transport, focusing for the first time on blood stasis in the LAA, using a high-fidelity patient-specific computational fluid dynamics (CFD) model. We use an Immersed Boundary Method, simulating the flow in a fixed, uniform Cartesian mesh and imposing the movement of the LA walls with a moving Lagrangian mesh generated from 4D Computerized Tomography images. We analyzed LA anatomies from eight patients with varying atrial function, including three with AF and either a LAA thrombus or a history of Transient Ischemic Attacks (TIAs). Using four different flow splits (60/40% and 55/45% through right and left PVs, even flow rate, and same velocity through each PV), we found that flow patterns are sensitive to PV flow split variations, particularly in planes parallel to the mitral valve. Changes in PV flow split also had a significant impact on blood stasis and could contribute to increased risk for thrombosis inside the LAA, particularly in patients with AF and previous LAA thrombus or a history of TIAs. Our study highlights the importance of considering patient-specific PV flow split variations when assessing LA hemodynamics and identifying patients at increased risk for thrombosis and stroke. This knowledge is relevant to planning clinical procedures such as AF ablation or the implementation of LAA occluders.

Deep learning networks in the segmentation of the left atrial appendage in 2D ultrasound: A comparative analysis

Left atrial appendage (LAA) is the major source of thromboembolism in patients with non-valvular atrial fibrillation. Currently, LAA occlusion can be offered as a treatment for these patients, obstructing the LAA through a percutaneously delivered device. Nevertheless, correct device sizing is a complex task, requiring manual analysis of medical images. This approach is sub-optimal, time-demanding, and highly variable between experts. Different solutions were proposed to improve intervention planning, but, no efficient solution is available to 2D ultrasound, which is the most used imaging modality for intervention planning and guidance. In this work, we studied the performance of recently proposed deep learning methods when applied for the LAA segmentation in 2D ultrasound. For that, it was created a 2D ultrasound database. Then, the performance of different deep learning methods, namely Unet, UnetR, AttUnet, TransAttUnet was assessed. All networks were compared using seven metrics: i) Dice coefficient; ii) Accuracy iii) Recall; iv) Specificity; v) Precision; vi) Hausdorff distance and vii) Average distance error. Overall, the results demonstrate the efficiency of AttUnet and TransAttUnet with dice scores of 88.62% and 89.28%, and accuracy of 88.25% and 86.30%, respectively. The current results demonstrate the feasibility of deep learning methods for LAA segmentation in 2D ultrasound.Clinical relevance- Our results proved the clinical potential of deep neural networks for the LAA anatomical analysis.

Centerline depth world for left atrial appendage orifice localization using reinforcement learning

Left atrial appendage (LAA) occlusion (LAAO) is a minimally invasive implant-based method to prevent cardiovascular stroke in patients with non-valvular atrial fibrillation. Assessing the LAA orifice in preoperative CT angiography plays a crucial role in choosing an appropriate LAAO implant size and a proper C-arm angulation. However, accurate orifice localization is hard because of the high anatomic variation of LAA, and unclear position and orientation of the orifice in available CT views. With the major research focus being on LAA segmentation, the only existing computational method for orifice localization utilized a rule-based decision. Nonetheless, using such a fixed rule may yield high localization error due to the varied anatomy of LAA. While deep learning-based models usually show improvements under such variation, learning an effective localization model is difficult because of the tiny orifice structure compared to the vast search space of CT volume. In this paper, we propose a centerline depth-based reinforcement learning (RL) world for effective orifice localization in a small search space. In our scheme, an RL agent observes the centerline-to-surface distance and navigates through the LAA centerline to localize the orifice. Thus, the search space is significantly reduced facilitating improved localization. The proposed formulation could result in high localization accuracy compared to the expert annotations. Moreover, the localization process takes about 7.3 s which is 18 times more efficient than the existing method. Therefore, this can be a useful aid to physicians during the preprocedural planning of LAAO.

Adversarial-based latent space alignment network for left atrial appendage segmentation in transesophageal echocardiography images

Left atrial appendage (LAA) is a leading cause of atrial fibrillation and thrombosis in cardiovascular disease. Clinicians can rely on LAA occlusion (LAAO) to effectively prevent and treat ischaemic strokes attributed to the LAA. The correct selection of the LAAO is one of the most critical stages in the successful surgical process, which relies on the quantification of the anatomical structure of the LAA for successful intervention in LAAO. In this paper, we propose an adversarial-based latent space alignment framework for LAA segmentation in transesophageal echocardiography (TEE) images by introducing prior knowledge from the label. The proposed method consists of an LAA segmentation network, a label reconstruction network, and a latent space alignment loss. To be specific, we first employ ConvNeXt as the backbone of the segmentation and reconstruction network to enhance the feature extraction capability of the encoder. The label reconstruction network then encodes the prior shape features from the LAA labels to the latent space. The latent space alignment loss consists of the adversarial-based alignment and the contrast learning losses. It can motivate the segmentation network to learn the prior shape features of the labels, thus improving the accuracy of LAA edge segmentation. The proposed method was evaluated on a TEE dataset including 1,783 images and the experimental results showed that the proposed method outperformed other state-of-the-art LAA segmentation methods with Dice coefficient, AUC, ACC, G-mean, and Kappa of 0.831, 0.917, 0.989, 0.911, and 0.825, respectively.

Stroke risk evaluation for patients with atrial fibrillation: Insights from left atrial appendage

Left atrial appendage (LAA) is believed to be a common site of thrombus formation in patients with atrial fibrillation (AF). However, the commonly-applied stroke risk stratification model (such as. CHA2DS2-VASc score) does not include any structural or hemodynamic features of LAA. Recent studies have suggested that it is important to incorporate LAA geometrical and hemodynamic features to evaluate the risk of thrombus formation in LAA, which may better delineate the AF patients for anticoagulant administration and prevent strokes. This review focuses on the LAA-related factors that may be associated with thrombus formation and cardioembolic events.

Feasibility and Accuracy of Automated Three-Dimensional Echocardiographic Analysis of Left Atrial Appendage for Transcatheter Closure

BACKGROUND

Procedural success of transcatheter left atrial appendage closure (LAAC) is dependent on correct device selection. Three-dimensional (3D) transesophageal echocardiography (TEE) is more accurate than the two-dimensional modality for evaluation of the complex anatomy of the left atrial appendage (LAA). However, 3D transesophageal echocardiographic analysis of the LAA is challenging and highly expertise dependent. The aim of this study was to evaluate the feasibility and accuracy of a novel software tool for automated 3D analysis of the LAA using 3D transesophageal echocardiographic data.

METHODS

Intraprocedural 3D transesophageal echocardiographic data from 158 patients who underwent LAAC were retrospectively analyzed using a novel automated LAA analysis software tool. On the basis of the 3D transesophageal echocardiographic data, the software semiautomatically segmented the 3D LAA structure, determined the device landing zone, and generated measurements of the landing zone dimensions and LAA length, allowing manual editing if necessary. The accuracy of LAA preimplantation anatomic measurement reproducibility and time for analysis of the automated software were compared against expert manual 3D analysis. The software feasibility to predict the optimal device size was directly compared with implanted models.

RESULTS

Automated 3D analysis of the LAA on 3D TEE was feasible in all patients. There was excellent agreement between automated and manual measurements of landing zone maximal diameter (bias, -0.32; limits of agreement, -3.56 to 2.92), area-derived mean diameter (bias, -0.24; limits of agreement, -3.12 to 2.64), and LAA depth (bias, 0.02; limits of agreement, -3.14 to 3.18). Automated 3D analysis, with manual editing if necessary, accurately identified the implanted device size in 90.5% of patients, outperforming two-dimensional TEE (68.9%; P < .01). The automated software showed results competitive against the manual analysis of 3D TEE, with higher intra- and interobserver reproducibility, and allowed quicker analysis (101.9 ± 9.3 vs 183.5 ± 42.7 sec, P < .001) compared with manual analysis.

CONCLUSIONS

Automated LAA analysis on the basis of 3D TEE is feasible and allows accurate, reproducible, and rapid device sizing decision for LAAC.

A comprehensive comparison of various patient-specific CFD models of the left atrium for atrial fibrillation patients

BACKGROUND

Recently, advances in medical imaging, segmentation techniques, and high-performance computing have supported the use of patient-specific computational fluid dynamics (CFD) simulations. At present, CFD-compatible atrium geometries can be easily reconstructed from atrium images, providing important insight into the atrial fibrillation (AF) phenomenon, and assistance during therapy selection and surgical procedures. However, the hypothesis assumed for such CFD models should be adequately validated.

AIM

This work aims to perform an extensive study of the different hypotheses that are commonly assumed when performing atrial simulations for AF patients, as well as to evaluate and compare the range of indices that are usually applied to assess thrombus formation within the left atrium appendage (LAA).

METHODS

The atrial geometries of two AF patients have been segmented. The resulting geometries have been registered and interpolated to construct a dynamic mesh, which has been employed to compare the rigid and flexible models. Two families of hemodynamic indices have been calculated and compared: wall shear-based and blood age distribution-based.

RESULTS

The findings of this study illustrate the importance of validating the rigid atrium hypothesis when utilizing an AF CFD model. In particular, the absence of the A-wave contraction does not avoid a certain degree of passive atrial contraction, making the rigid model a poor approximation in some cases. Moreover, a new thrombosis predicting index has been proposed, i.e., M4, which has been shown to predict stasis more effectively than other indicators.

Use of radiomics to differentiate left atrial appendage thrombi and mixing artifacts on single-phase CT angiography

To assess if radiomics can differentiate left atrial appendage (LAA) contrast-mixing artifacts and thrombi on early-phase CT angiography without the need for late-phase images. Our study included 111 patients who underwent early- and late-phase, contrast-enhanced cardiac CT. Of these, 79 patients had LAA filling defects from thrombus (n = 46, mean age: 72  ±  12 years, M:F 26:20) or contrast-mixing artifact (n = 33, mean age: 71  ±  13 years, M:F 21:12) on early-contrast-enhanced phase. The remaining 32 patients (mean age: 66  ±  10 years, M:F 19:13) had homogeneous LAA opacification without filling defects. The entire LAA volume on early-phase CT images was manually segmented to obtain radiomic features (Frontier, Siemens). A radiologist assessed for the presence of LAA filling defects and recorded the size and mean CT attenuation (HU) of filling defects and normal LAA. The data were analyzed using multiple logistic regression with receiver operating characteristics area under the curve (AUC) as an output. The radiologist correctly identified all 32 patients without LAA filling defects, 42/46 LAA with thrombi, and 23/33 contrast mixing artifacts. Although HU of LAA thrombi and contrast mixing artifacts was significantly different, with the lowest AUC (0.66), it was inferior to both radiologist assessment and radiomics (p = 0.05). Combination of radiologist assessment and radiomics (AUC 0.92) was superior to HU (0.66), radiomics (0.85), and radiologist (0.80) alone (p < 0.008). Radiomics can differentiate between LAA filling defects from thrombi and contrast mixing artifacts on early-phase contrast-enhanced CT images without the need for late-phase CT.