Back

Optimisation of asymmetric field pulses for transcranial magnetic stimulation

Ma, K.; Goetz, S. M.

2023-05-01 neuroscience
10.1101/2023.05.01.538933 bioRxiv
Show abstract

Transcranial magnetic stimulation (TMS) is a widely-used noninvasive brain stimulation technique through electromagnetic induction. Nowadays commercial TMS devices routinely use conventional biphasic pulses for repetitive TMS protocols and monophasic pulses for single-pulse stimulation. They respectively generate underdamped or damped cosinusoidal electric field pulses that have been proven to be power-inefficient. Recently, symmetric field pulses with near-rectangular electric fields show great potential in terms of energy loss and coil heating, but only limited studies have investigated asymmetric field pulses with different asymmetry levels for the induced electric field waveforms. Thus, this paper optimises and searches a wide range of potential current waveforms with the goal of minimising energy loss and coil heating. The optimised results demonstrated that asymmetric field pulses with near-rectangular electric fields have significantly lower power consumption than conventional ones. These optimised waveforms commonly consist of an initial falling phase followed by rapidly rising and falling phases, trailed by a slow decay to zero. Interestingly, the initial phase has a decay time constant around 260 {micro}s and introduces a pulse-duration-dependent negative bias for the current baseline to minimise the energy loss and coil heating. Thus, it is possible to directly design asymmetric field pulses with various asymmetry ratios by using several prediction equations rather than running optimisation. These results also suggest that introducing such an initial current phase could likely significantly reduce the coil heating of most TMS pulse shapes to improve their power efficiencies.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.