Convolution model for gradient-induced cardiac stimulation risk assessment

Motivation: Accurate cardiac stimulation (CS) risk assessment is important for modern high-performance gradient coils but evaluation methods for arbitrary waveforms are currently lacking. Goal(s): To develop a stimulation response (R) calculation method for CS that is analogous to PNS evaluation currently recognized by international standards. Approach: Both exponential and hyperbolic functional models were used to construct a CS convolution kernel, and R values were calculated on representative neuroimaging sequences. Results: Exponential model produced higher (more conservative) maximum R values than the hyperbolic model. Compared to PNS, a longer chronaxie time made CS less sensitive to rapid bipolar waveforms such as EPI. Impact: We extended sequence-specific PNS risk evaluation to cardiac stimulation. Our work enables systematic evaluation of CS risk for arbitrary gradient waveforms, and can help ensure patient safety in high-performance gradient systems.