Interpolation of electrophysiology parameters using OpenEP: technology development and clinical application. (19th May 2022)

• Record Type: Journal Article
• Authors: Williams, S; Roney, CH; Connolly, A; Smith, P; Bishop, M; Niederer, S; Whitaker, J; Corrado, C; Kotadia, I; O'hare, D; Fitzpatrick, N; Sim, I; O'neill, M
• Is Part Of: Europace. Volume 24:Supplement 1(2022)
• Publication Date: 2022-05-19
• Subjects: Arrhythmia -- Treatment -- Periodicals; Cardiac pacing -- Periodicals; Catheter ablation -- Periodicals; Heart -- Physiology -- Periodicals; Electrophysiology -- Periodicals; 617.4120645
• DOI: 10.1093/europace/euac053.049
• Languages: English
• ISSNs: 1099-5129

Abstract

Abstract: Funding Acknowledgements: Type of funding sources: Foundation. Main funding source(s): British Heart Foundation Background: Interpolation of data is common during clinical electrophysiology procedures. Applications include local activation mapping, voltage mapping and novel techniques including Sparkle and Coherence mapping. Nevertheless, underlying interpolation algorithms are proprietary and therefore challenging to reproduce. Importantly, direct comparison of electroanatomic datasets between system vendors is therefore not possible. Purpose: We sought to (1) develop an open-source architecture for interpolation within the Open Electrophysiology Framework for Research (OpenEP; https://openep.io ); (2) to provide three interpolation methods within this architecture and (3) to evaluate their performance against clinical data. Method: The software architecture is shown in Figure 1A. The currently available methods are Radial Basis [1], Scattered Interpolant [2] and Local Smoothing [3]. Default parameters for each method are shown in Figure 1B. The performance of each method was assessed using clinical left atrial mapping data, using the default options for each scheme. Following interpolation, a sample of 1000 activation/voltage points per mesh was used for analysis. For each interpolation method, correlation with clinical data was assessed using the intra-class correlation coefficient, whilst agreement was assessed using Bland Altman limits of agreement. Results: For activation mapping, radial basis interpolation resulted in a smoother field than local smoothing, whilst scattered interpolation required more filtering of extreme values. Correlations between interpolated and original activation times were excellent for all interpolation schemes (radial basis R=0.91, p<0.0001; local smoothing R=0.95, p<0.0001; scattered interpolant R=0.92, p<0.0001). Local smoothing resulted in the narrowest 95 percent limits of agreement (-19 to +20ms), compared to radial basis (-24 to +28ms) and scattered interpolation (-22 to +25ms). For voltage mapping, the interpolation schemes resulted in similar appearances of low voltage areas, however correlations with clinical data were weaker than for activation mapping (radial basis R=0.84, p<0.0001; local smoothing R=0.82, p<0.0001; scattered interpolant R=0.79, p<0.0001). The 95 percent limits of agreement were wide as a proportion of the mean data values, ranging from 83% (-0.8 to +0.66mV) for local smoothing to 97% (-0.78 to +0.63mV) for radial basis interpolation. Conclusion: An extensible architecture is provided for data interpolation in OpenEP together with three interpolation methods. The methods performed wellfor local activation time interpolation but variation compared to clinical data was greater for voltage mapping. This new architecture will permit the optimisation of interpolation methods against "gold standard" simulation or histological data and facilitate comparison of datasets between system vendors.