Assessment of a Threshold Method for Computing Activation Maps from Reconstructed Transmembrane Voltages

TL;DR

This paper introduces a simple threshold-based approach for generating activation maps from reconstructed transmembrane voltages, which effectively reduces artifacts and preserves true conduction discontinuities in cardiac imaging.

Contribution

A novel threshold method for activation map computation is proposed, offering a straightforward alternative to complex smoothing algorithms in electrocardiographic imaging.

Findings

Threshold method produces smooth activation maps with true lines of block.

Combining threshold method with Tikhonov regularization enhances map quality.

Performance variability highlights challenges in universal post-processing solutions.

Abstract

Electrocardiographic imaging non-invasively reconstructs activation maps of the heart from temporal body surface potential maps by post-processing solutions of an inverse problem. Typically, activation times are detected through the maximal deflection of the temporal or spatial derivative of recovered extracellular or transmembrane potentials. However, this method can introduce artificial lines of block in the map, falsely indicating a pathology. Consequently, several complex algorithms have been developed in an attempt to smooth activation maps while preserving true discontinuities. We propose a straightforward method for computing activation maps from recovered transmembrane voltages, wherein activation time is defined as the first time a predefined threshold is crossed. We evaluate this thresholdbased method against traditional deflection-based methods using simulated data, following the approach of Shuler et al. [1]. Our findings indicate that the threshold method, when combined with classical Tikhonov regularization, produces smooth activation maps even in the presence of true lines of block. Given the variability in performance of deflectionbased methods compared to those reported in [1], we emphasize the difficulty of establishing a universally effective post-processing method for computing activation maps.