Dataset Information

Atrial fibrillation ablation outcome prediction with a machine learning fusion framework incorporating cardiac computed tomography.

ABSTRACT:

Background

Structural changes in the left atrium (LA) modestly predict outcomes in patients undergoing catheter ablation for atrial fibrillation (AF). Machine learning (ML) is a promising approach to personalize AF management strategies and improve predictive risk models after catheter ablation by integrating atrial geometry from cardiac computed tomography (CT) scans and patient-specific clinical data. We hypothesized that ML approaches based on a patient's specific data can identify responders to AF ablation.

Methods

Consecutive patients undergoing AF ablation, who had preprocedural CT scans, demographics, and 1-year follow-up data, were included in the study for a retrospective analysis. The inputs of models were CT-derived morphological features from left atrial segmentation (including the shape, volume of the LA, LA appendage, and pulmonary vein ostia) along with deep features learned directly from raw CT images, and clinical data. These were merged intelligently in a framework to learn their individual importance and produce the optimal classification.

Results

Three hundred twenty-one patients (64.2 ± 10.6 years, 69% male, 40% paroxysmal AF) were analyzed. Post 10-fold nested cross-validation, the model trained to intelligently merge and learn appropriate weights for clinical, morphological, and imaging data (AUC 0.821) outperformed those trained solely on clinical data (AUC 0.626), morphological (AUC 0.659), or imaging data (AUC 0.764).

Conclusion

Our ML approach provides an end-to-end automated technique to predict AF ablation outcomes using deep learning from CT images, derived structural properties of LA, augmented by incorporation of clinical data in a merged ML framework. This can help develop personalized strategies for patient selection in invasive management of AF.

SUBMITTER: Razeghi O

PROVIDER: S-EPMC10857794 | biostudies-literature | 2023 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Atrial fibrillation ablation outcome prediction with a machine learning fusion framework incorporating cardiac computed tomography.

Razeghi Orod O Kapoor Ridhima R Alhusseini Mahmood I MI Fazal Muhammad M Tang Siyi S Roney Caroline H CH Rogers Albert J AJ Lee Anson A Wang Paul J PJ Clopton Paul P Rubin Daniel L DL Narayan Sanjiv M SM Niederer Steven S Baykaner Tina T

Journal of cardiovascular electrophysiology 20230427 5

<h4>Background</h4>Structural changes in the left atrium (LA) modestly predict outcomes in patients undergoing catheter ablation for atrial fibrillation (AF). Machine learning (ML) is a promising approach to personalize AF management strategies and improve predictive risk models after catheter ablation by integrating atrial geometry from cardiac computed tomography (CT) scans and patient-specific clinical data. We hypothesized that ML approaches based on a patient's specific data can identify re ...[more]

PMID: 36934383

Similar Datasets

Project description:BackgroundThe significance of the right atrial appendage (RAA) and right atrium (RA) in the recurrence of atrial fibrillation (AF) after radiofrequency ablation (RFA) remains uncertain. This retrospective case-control study aimed to quantitatively evaluate the role of morphological parameters of the RAA and RA in the recurrence of AF after RFA based on 256-slice spiral computed tomography (CT).MethodsA total of 297 patients with AF who underwent RFA for the first time between January 1 and October 31, 2020, were enrolled in the study, and they were divided into a nonrecurrence group (n=214) and a recurrence group (n=83). The volume of the RA, RAA and left atrium (LA); height of the RAA; long and short diameter, perimeter, and area of the RAA base; right atrial anteroposterior diameter; tricuspid annulus diameter; crista terminalis thickness; and cavotricuspid isthmus (CVTI) were measured, and the clinical data of patients were collected.Results(I) Multivariable logistic regression analysis followed by univariable logistic regression analysis showed that the height of the RAA [odds ratio (OR) =1.124; 95% confidence interval (CI): 1.024-1.233; P=0.014], short diameter of the RAA base (OR =1.247; 95% CI: 1.118-1.391; P=0.001), crista terminalis thickness (OR =1.594; 95% CI: 1.052-2.415; P=0.028) and duration of AF (OR =1.009; 95% CI: 1.003-1.016; P=0.006) were independent predictors of postradiofrequency ablation AF recurrence. (II) Receiver operating characteristic (ROC) curve analysis showed that the prediction model constructed according to the multivariate logistic regression analysis presented good accuracy [area under the curve (AUC) =0.840; P=0.001]. A short diameter of the RAA base >26.95 mm had the highest predictive value for AF recurrence, with a sensitivity of 0.614 and a specificity of 0.822 (AUC =0.786, P=0.001). Pearson correlation analysis showed that there was a significant correlation between right atrial volume and left atrial volume (r=0.720, P<0.001).ConclusionsA significant increase in diameter and volume of the RAA and RA and tricuspid annulus diameter may correlate with postradiofrequency ablation AF recurrence. The height of the RAA, short diameter of the RAA base, crista terminalis thickness, and AF duration were independent predictors of recurrence. Among them, the short diameter of the RAA base had the highest predictive value for recurrence.

Project description:BackgroundMachine learning is a promising approach to personalize atrial fibrillation management strategies for patients after catheter ablation. Prior atrial fibrillation ablation outcome prediction studies applied classical machine learning methods to hand-crafted clinical scores, and none have leveraged intracardiac electrograms or 12-lead surface electrocardiograms for outcome prediction. We hypothesized that (1) machine learning models trained on electrograms or electrocardiogram (ECG) signals can perform better at predicting patient outcomes after atrial fibrillation ablation than existing clinical scores and (2) multimodal fusion of electrogram, ECG, and clinical features can further improve the prediction of patient outcomes.MethodsConsecutive patients who underwent catheter ablation between 2015 and 2017 with panoramic left atrial electrogram before ablation and clinical follow-up for at least 1 year following ablation were included. Convolutional neural network and a novel multimodal fusion framework were developed for predicting 1-year atrial fibrillation recurrence after catheter ablation from electrogram, ECG signals, and clinical features. The models were trained and validated using 10-fold cross-validation on patient-level splits.ResultsOne hundred fifty-six patients (64.5±10.5 years, 74% male, 42% paroxysmal) were analyzed. Using electrogram signals alone, the convolutional neural network achieved an area under the receiver operating characteristics curve (AUROC) of 0.731, outperforming the existing APPLE scores (AUROC=0.644) and CHA2DS2-VASc scores (AUROC=0.650). Similarly using 12-lead ECG alone, the convolutional neural network achieved an AUROC of 0.767. Combining electrogram, ECG, and clinical features, the fusion model achieved an AUROC of 0.859, outperforming single and dual modality models.ConclusionsDeep neural networks trained on electrogram or ECG signals improved the prediction of catheter ablation outcome compared with existing clinical scores, and fusion of electrogram, ECG, and clinical features further improved the prediction. This suggests the promise of using machine learning to help treatment planning for patients after catheter ablation.

Project description:BackgroundThere is an increasing evidence that pulmonary vein (PV) enlargement is associated with atrial fibrillation (AF); however, the predictive value of PV enlargement in AF recurrence remains unclear. This study sought to evaluate whether PV volume quantification derived from cardiac computed tomographic angiography (CCTA) could serve as a predictive indicator of the success of the catheter ablation (CA) procedure.MethodsThe data of 160 patients diagnosed with AF who underwent both CCTA and CA treatments from January to June 2020 were retrospectively examined; the CCTA was conducted before the CA surgery. The study focused on documenting the PV structure, and the volume of the PV and left atrium (LA). The clinical, CCTA, and echocardiographic predictors of the recurrence and no-recurrence groups were compared. A multivariable logistic regression analysis was performed to adjust for confounders. Receiver operating characteristic (ROC) curves were analyzed to assess the predictive performance of the predictors of AF recurrence.ResultsOf the 160 patients [55.6% male, 62.00 (55.25-68.00) years, 23.1% with persistent AF], 45 (28.1%) experienced AF recurrence within a one-year period. Notably, patients with AF recurrence had elevated CHADS2 scores (P=0.020) and increased LA and PV volumes (P<0.05). Patients with persistent AF (n=37) had significantly larger LA volume indexes (P<0.001) than those with paroxysmal AF, but there was no difference between the two groups in terms of the PV maximum volume index (P=0.200). Moreover, the PV maximum volume index [odds ratio (OR): 1.244, 95% confidence interval (CI): 1.008-1.536, P=0.042] and the LA minimum volume index (OR: 1.026, 95% CI: 1.001-1.052, P=0.038) were found to be significant predictors of AF recurrence. The ROC curves revealed that the PV maximum volume index threshold for predicting AF recurrence was 7.13 mL/m2, with a sensitivity of 84.4% and a specificity of 34.8% [area under the curve (AUC): 0.635, 95% CI: 0.540-0.730, P=0.008], and the LA minimum volume index threshold for predicting AF recurrence was 46.16 mL/m2, with a sensitivity of 88.9% and a specificity of 31.3% (AUC: 0.629, 95% CI: 0.534-0.723, P=0.012). A sub-analysis of patients with a lower left atrial dimension (LAD ≤38 mm in females, LAD ≤40 mm in males, n=120) demonstrated that the PV maximum volume index was a noteworthy indicator of AF recurrence (OR: 1.443: 95% CI: 1.145-1.820, P=0.002). Conversely, no significant correlation between AF recurrence and the LA volume index was found. The AUC value for the PV maximum volume index predictive of recurrent AF was 0.680 (95% CI: 0.577-0.781, P=0.003), with a sensitivity of 75.8%, specificity of 54%, and the cut-off value of the maximum AUC was 7.89 mL/m2.ConclusionsPV volume, derived from CCTA, may help to predict the recurrence of AF after CA, and is superior to the LA size in patients with less pronounced LA enlargement.

Project description:BackgroundRadiofrequency catheter ablation (RFCA) represents an important treatment option for atrial fibrillation (AF); however, the recurrence rate following surgery is relatively high. This study aimed to predict the recurrence of AF after RFCA using interpretable machine learning models that combined the triglyceride-glucose (TyG) index and the quantification of left atrial epicardial and pericoronary adipose tissue.MethodsThis retrospective study included 325 patients with AF who underwent their first successful RFCA, among whom 79 had confirmed recurrence. The preoperative clinical data of patients were collected, the TyG index was calculated, and computed tomography (CT) image features were quantitatively measured. Multivariate Cox regression analysis was used to identify the independent risk factors for RFCA recurrence, and adjustments being made for various confounding factors. Post-hoc subgroup analysis was conducted to evaluate the predictive value of the TyG index for recurrence in different patient subgroups. Prediction models based on six machine learning algorithms were constructed. The optimal model's features were evaluated using Shapley additive explanations (SHAP).ResultsAfter adjustment were made for various confounding factors such as comorbidities of AF, Cox regression showed that the volume of left atrial epicardial adipose tissue (LA-EAT), LA-EAT attenuation, left circumflex coronary artery fat attenuation index (LCX-FAI), and the TyG index were independent risk factors for recurrence after RFCA (P<0.001). The support vector machine (SVM) model based on these combined indicators had the best predictive performance, with an area under the curve of 0.793 [95% confidence interval (CI): 0.782-0.805] in the validation set, while its accuracy and positive predictive value were 0.804 and 0.710, respectively. The predictive efficiency of the TyG index appeared to be independent of type 2 diabetes mellitus (T2DM) status (Pinteraction=0.660).ConclusionsThe SVM model that integrated the TyG index and quantitative CT imaging variables demonstrated good predictive ability for post-RFCA recurrence in patients with AF. Furthermore, the TyG index appeared capable of predicting recurrence independently of T2DM status.

Project description:Radiofrequency ablation (RFA) is the treatment of choice for atrial fibrillation (AF). Additionally, the utilization of 3D printing for cardiac models offers an in-depth insight into cardiac anatomy and cardiovascular diseases. The study aims to evaluate the clinical utility and outcomes of RFA following in vitro visualization of the left atrium (LA) and pulmonary vein (PV) structures via 3D printing (3DP). Between November 2017 and April 2021, patients who underwent RFA at the First Affiliated Hospital of Xinxiang Medical University were consecutively enrolled and randomly allocated into two groups: the 3DP group and the control group, in a 1:1 ratio. Computed tomography angiography (CTA) was employed to capture the morphology and diameter of the LA and PV, which facilitated the construction of a 3D entity model. Additionally, surgical procedures were simulated using the 3D model. Parameters such as the duration of the procedure, complications, and rates of RFA recurrence were meticulously documented. Statistical analysis was performed using the t-test or Mann-Whitney U test to evaluate the differences between the groups, with a P-value of less than 0.05 considered statistically significant. In this study, a total of 122 patients were included, with 53 allocated to the 3DP group and 69 to the control group. The analysis of the morphological measurements of the LA and PV taken from the workstation or direct entity measurement showed no significant difference between the two groups (P > 0.05). However, patients in the 3DP group experienced significantly shorter RFA times (97.03 ± 28.39 compared to 120.51 ± 44.76 min, t = 3.05, P = 0.003), reduced duration of radiation exposure (2.55 [interquartile range 2.01, 3.24] versus 3.20 [2.28, 3.91] min, Z = 3.23, P < 0.001), and shorter modeling times (7.68 ± 1.03 compared to 8.89 ± 1.45 min, t = 5.38, P < 0.001). 3DP technology has the potential to enhance standard RFA practices by reducing the time required for intraoperative interventions and exposure to radiation.

Project description:Catheter ablation is a curative therapeutic approach for atrial fibrillation (AF). Ablation of rotational sources based on basket catheter measurements has been proposed as a promising approach in patients with persistent AF to complement pulmonary vein isolation. However, clinically reported success rates are equivocal calling for a mechanistic investigation under controlled conditions. We present a computational framework to benchmark ablation strategies considering the whole cycle from excitation propagation to electrogram acquisition and processing to virtual therapy. Fibrillation was induced in a patient-specific 3D volumetric model of the left atrium, which was homogeneously remodeled to sustain reentry. The resulting extracellular potential field was sampled using models of grid catheters as well as realistically deformed basket catheters considering the specific atrial anatomy. The virtual electrograms were processed to compute phase singularity density maps to target rotor tips with up to three circular ablations. Stable rotors were successfully induced in different regions of the homogeneously remodeled atrium showing that rotors are not constrained to unique anatomical structures or locations. Density maps of rotor tip trajectories correctly identified and located the rotors (deviation < 10 mm) based on catheter recordings only for sufficient resolution (inter-electrode distance ≤3 mm) and proximity to the wall (≤10 mm). Targeting rotor sites with ablation did not stop reentries in the homogeneously remodeled atria independent from lesion size (1-7 mm radius), from linearly connecting lesions with anatomical obstacles, and from the number of rotors targeted sequentially (≤3). Our results show that phase maps derived from intracardiac electrograms can be a powerful tool to map atrial activation patterns, yet they can also be misleading due to inaccurate localization of the rotor tip depending on electrode resolution and distance to the wall. This should be considered to avoid ablating regions that are in fact free of rotor sources of AF. In our experience, ablation of rotor sites was not successful to stop fibrillation. Our comprehensive simulation framework provides the means to holistically benchmark ablation strategies in silico under consideration of all steps involved in electrogram-based therapy and, in future, could be used to study more heterogeneously remodeled disease states as well.

Dataset Information

Atrial fibrillation ablation outcome prediction with a machine learning fusion framework incorporating cardiac computed tomography.

Background

Methods

Results

Conclusion

Publications

Atrial fibrillation ablation outcome prediction with a machine learning fusion framework incorporating cardiac computed tomography.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets