Abstract
Non-invasive brain stimulation (NIBS) techniques using non-ionizing electromagnetic fields face a persistent trade-off between spatial focality and stimulation depth when targeting deep brain structures. This study systematically compares two temporal interference (TI) stimulation modalities-current-injection-based TI transcranial alternating current stimulation (TI-tACS) and electromagnetic-induction-based TI transcranial magnetic stimulation (TI-TMS)-with respect to their electric field distributions and depth-attenuation profiles at a generic deep target positioned at a depth consistent with the approximate location of the hippocampus. A four-layer concentric spherical finite element head model comprising scalp, skull, cerebrospinal fluid, and brain tissue was constructed. Under a unified simulation framework, the envelope electric field intensity distributions, shallow-to-deep energy ratios, and radial attenuation behaviours of both modalities were quantitatively evaluated across a range of inter-coil/inter-electrode angles. Simulation results demonstrate that TI-tACS exhibits pronounced depth-dependent attenuation, with this phenomenon being mainly attributed to the high impedance of the skull. Deep-field enhancement in TI-tACS is consistently accompanied by increased superficial field exposure, and its spatial distribution is also sensitive to stimulation configuration parameters. In contrast, TI-TMS generates induced electric fields that are less influenced by variations in tissue conductivity, thereby yielding a relatively uniform depth-distribution profile. Under the present modelling conditions, TI-TMS demonstrates higher deep-targeting efficiency relative to scalp exposure, though at the cost of greater whole-brain field distribution. These complementary metrics suggest that the relative merits of the two modalities are application-dependent rather than absolute. These findings clarify the mechanistic differences between the two TI modalities with respect to electromagnetic field propagation and interaction with biological tissue, offering a computational basis for parameter optimisation in non-invasive deep brain stimulation research.