Dataset Information

Optical Mapping-Validated Machine Learning Improves Atrial Fibrillation Driver Detection by Multi-Electrode Mapping.

ABSTRACT:

Background

Atrial fibrillation (AF) can be maintained by localized intramural reentrant drivers. However, AF driver detection by clinical surface-only multielectrode mapping (MEM) has relied on subjective interpretation of activation maps. We hypothesized that application of machine learning to electrogram frequency spectra may accurately automate driver detection by MEM and add some objectivity to the interpretation of MEM findings.

Methods

Temporally and spatially stable single AF drivers were mapped simultaneously in explanted human atria (n=11) by subsurface near-infrared optical mapping (NIOM; 0.3 mm² resolution) and 64-electrode MEM (higher density or lower density with 3 and 9 mm² resolution, respectively). Unipolar MEM and NIOM recordings were processed by Fourier transform analysis into 28 407 total Fourier spectra. Thirty-five features for machine learning were extracted from each Fourier spectrum.

Results

Targeted driver ablation and NIOM activation maps efficiently defined the center and periphery of AF driver preferential tracks and provided validated annotations for driver versus nondriver electrodes in MEM arrays. Compared with analysis of single electrogram frequency features, averaging the features from each of the 8 neighboring electrodes, significantly improved classification of AF driver electrograms. The classification metrics increased when less strict annotation, including driver periphery electrodes, were added to driver center annotation. Notably, f1-score for the binary classification of higher-density catheter data set was significantly higher than that of lower-density catheter (0.81±0.02 versus 0.66±0.04, P<0.05). The trained algorithm correctly highlighted 86% of driver regions with higher density but only 80% with lower-density MEM arrays (81% for lower-density+higher-density arrays together).

Conclusions

The machine learning model pretrained on Fourier spectrum features allows efficient classification of electrograms recordings as AF driver or nondriver compared with the NIOM gold-standard. Future application of NIOM-validated machine learning approach may improve the accuracy of AF driver detection for targeted ablation treatment in patients.

SUBMITTER: Zolotarev AM

PROVIDER: S-EPMC7577986 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundVisual cortical prostheses (VCPs) have the potential to restore visual function to patients with acquired blindness. Successful implementation of VCPs requires the ability to reliably map the location of the phosphene produced by stimulation of each implanted electrode.ObjectiveTo evaluate the efficacy of different approaches to phosphene mapping and propose simple improvements to mapping strategy.MethodsWe stimulated electrodes implanted in the visual cortex of five blind and fifteen sighted patients. We tested two fixation strategies, unimanual fixation, where subjects placed a single index finger on a tactile fixation point and bimanual fixation, where subjects overlaid their right index finger over their left on the tactile point. In addition, we compared absolute mapping in which a single electrode was stimulated on each trial, and relative mapping with sequences containing stimulation of three to five phosphenes on each trial. Trial-to-trial variability present in relative mapping sequences was quantified.ResultsPhosphene mapping was less precise in blind subjects than in sighted subjects (2DRMS, 16 ± 2.9° vs. 1.9 ± 0.93°; t (18) = 18, p = <0.001). Within blind subjects, bimanual fixation resulted in more consistent phosphene localization than unimanual fixation (BS1: 4.0 ± 2.6° vs. 19 ± 4.7°, t (79) = 24, p < 0.001; BS2 4.1 ± 2.0° vs. 12 ± 2.7°, t (65) = 19, p < 0.001). Multi-point relative mapping had similar baseline precision to absolute mapping (BS1: 4.7 ± 2.6° vs. 3.9 ± 2.0°; BS2: 4.1 ± 2.0° vs. 3.2 ± 1.1°) but improved significantly when trial-to-trial translational variability was removed. Although multi-point mapping methods did reveal more of the functional organization expected in early visual cortex, subjects tended to artificially regularize the spacing between phosphenes. We attempt to address this issue by fitting a standard logarithmic map to relative multi-point sequences.ConclusionsRelative mapping methods, combined with bimanual fixation, resulted in the most precise estimates of phosphene organization. These techniques, combined with use of a standard logarithmic model of visual cortex, may provide a practical way to improve the implementation of a VCP.