Noninvasive Assessment of Atrial Fibrillation Complexity in Relation to Ablation Characteristics and Outcome

Abstract

Background: The use of surface recordings to assess atrial fibrillation (AF) complexity is still limited in clinical practice. We propose a noninvasive tool to quantify AF complexity from body surface potential maps (BSPMs) that could be used to choose patients who are eligible for AF ablation and assess therapy impact.Methods: BSPMs (mean duration: 7 ± 4 s) were recorded with a 252-lead vest in 97 persistent AF patients (80 male, 64 ± 11 years, duration 9.6 ± 10.4 months) before undergoing catheter ablation. Baseline cycle length (CL) was measured in the left atrial appendage. The procedural endpoint was AF termination. The ablation strategy impact was defined in terms of number of regions ablated, radiofrequency delivery time to achieve AF termination, and acute outcome. The atrial fibrillatory wave signal extracted from BSPMs was divided in 0.5-s consecutive segments, each projected on a 3D subspace determined through principal component analysis (PCA) in the current frame. We introduced the nondipolar component index (NDI) that quantifies the fraction of energy retained after subtracting an equivalent PCA dipolar approximation of heart electrical activity. AF complexity was assessed by the NDI averaged over the entire recording and compared to ablation strategy.Results: AF terminated in 77 patients (79%), whose baseline AF CL was 177 ± 40 ms, whereas it was 157 ± 26 ms in patients with unsuccessful ablation outcome (p = 0.0586). Mean radiofrequency emission duration was 35 ± 21 min; 4 ± 2 regions were targeted. Long-lasting AF patients (≥12 months) exhibited higher complexity, with higher NDI values (≥12 months: 0.12 ± 0.04 vs. <12 months: 0.09 ± 0.03, p < 0.01) and short CLs (<160 ms: 0.12 ± 0.03 vs. between 160 and 180 ms: 0.10 ± 0.03 vs. >180 ms: 0.09 ± 0.03, p < 0.01). More organized AF as measured by lower NDI was associated with successful ablation outcome (termination: 0.10 ± 0.03 vs. no termination: 0.12 ± 0.04, p < 0.01), shorter procedures (<30 min: 0.09 ± 0.04 vs. ≥30 min: 0.11 ± 0.03, p < 0.001) and fewer ablation targets (<4: 0.09 ± 0.03 vs. ≥4: 0.11 ± 0.04, p < 0.01).Conclusions: AF complexity can be noninvasively quantified by PCA in BSPMs and correlates with ablation outcome and AF pathophysiology.

Noninvasive One-Year Ablation Outcome Prediction for Paroxysmal Atrial Fibrillation Using Trajectories of Activation from Body Surface Potential Maps

“…As one signal contains multiple cycles of atrial activity (see Fig 4), segmentation into AF cycles is essential for cycle-to-cycle comparison. Since the activated patch size reflects the strength of the source, its local minimum was used as a marker to segment the trajectory into AF cycles and obtain trajectories of cycles (ToCs), with the constraint that the peak-to-peak interval was ≥ 90ms [2] and ≤ 350ms. A resulting segmentation is shown in Fig 2 and Fig 4. Predicting AF recurrence based on patient ToCs forms a multiple-instance classification problem [5], in which a varying number of ToCs was recorded for each patient were used to predict their one-year clinical outcome.…”

“…It was sought to determine whether the analysis of these signals could predict whether or not patients remain AF-free after CA at a one-year horizon. Existing markers to predict immediate CA success include AF cycle length (CL); nondipolar component index (NDI), a predictor for termination on persistent AF using BSPM [2]; and spatial variability (SPV), a predictor for short-term ablation outcome on persistent AF using 12-lead ECG [3]. However, none of these methods explicitly takes into account the spatial-temporal movement of the electrical activated region of atria.…”

Atrial Fibrillation Spatiotemporal Complexity Is Affected by Pulmonary Vein Isolation

“…BSPMs were acquired with a 252-lead vest (ECVUE, Cardioinsight Technologies) at 1 kHz before and after PVI. The multilead atrial fibrillatory wave (f-wave) signal (9±6 s) was processed as in [16], yielding a L × N matrix Y = [y(1) . .…”

“…Changes in the dominant frequency are also detectable on the standard electrocardiogram in response to PVI [15], although with some variability across the leads. More generally, AF complexity quantified by principal component analysis (PCA) of body surface potential maps (BSPMs) correlates with CA duration and strategy and is predictive of its outcome [16]. This study aims to investigate to which extent PVI affects AF organization measured from BSPMs.…”

“…This study aims to investigate to which extent PVI affects AF organization measured from BSPMs. Variations in signal features determined by PCA as in [16,17] were linked to changes in AF complexity occurring during PVI. Our approach provides deeper insights into the efficacy and the relevance of this intervention to persistent AF treatment.…”