The inverse reconstruction of cardiac electrical activity from body surface potentials (BSP) is an ill-posed problem where small levels of error can have a large impact on solutions. While the impact of signal noise on inverse reconstructions has been studied, to date there has been no investigation of the impact of poor BSP electrode contact.
This study examines this using potential (MFS) and activation (EDL) based inverse methods. Inverse reconstructions were performed on simulated single site pacing data with a single BSP in poor contact. To simulate poor contact, a simple model of reducing BSP amplitudes by 20%, 40%, 60%, 80% and 100%, was used, and this was performed for all 163 BSPs. ‘Gold standard’ solutions were defined as reconstructions without amplitude reduction.
For the MFS, reduced BSP amplitude near the heart severely impacted reconstruction accuracy. That is, small areas of the heart demonstrated electrogram inverse (CC<0%) and activation errors of 31-69 ms with a 40% amplitude reduction, and 39-117 ms with a 60% reduction. The size of the area affected was related to the degree of amplitude reduction and the location to the position of the affected BSP.
The EDL method was more robust than the MFS. That is the general activation pattern remained the same, and errors in activation were smaller than with the MFS (9-41 ms with 60% amplitude reduction). The area affected with large activation errors (>40 ms) was much smaller than with the MFS (0-3% vs 1-24% with 60% reduction). However, where these larger activation errors occur, artificial regions of slowed conduction appeared.
Impaired torso electrode contact introduces drastic and subtle errors into inverse solutions for both potential- and activation-based methods. These errors could result in misdiagnosis and subsequent treatment of patients. Methods to ensure good electrode contact should be developed and implemented.