Real-time estimation of electric fields induced by transcranial magnetic stimulation with deep neural networks

Abstract

Background: Transcranial magnetic stimulation (TMS) plays an important role in treatment of mental and neurological illnesses, and neurosurgery. However, it is difficult to target specific brain regions accurately because the complex anatomy of the brain substantially affects the shape and strength of the electric fields induced by the TMS coil. A volume conductor model can be used for determining the accurate electric fields; however, the construction of subject-specific anatomical head structures is timeconsuming. Objective: The aim of this study is to propose a method to estimate electric fields induced by TMS from only T1 magnetic resonance (MR) images, without constructing a subject-specific anatomical model. Methods: Very large sets of electric fields in the brain of subject-specific anatomical models, which are constructed from T1 and T2 MR images, are computed by a volume conductor model. The relation between electric field distribution and T1 MR images is used for machine learning. Deep neural network (DNN) models are applied for the first time to electric field estimation. Results: By determining the relationships between the T1 MR images and electric fields by DNN models, the process of electric field estimation is markedly accelerated (to 0.03 s) due to the absence of a requirement for anatomical head structure reconstruction and volume conductor computation. Validation shows promising estimation accuracy, and rapid computations of the DNN model are apt for practical applications. Conclusion: The study showed that the DNN model can estimate the electric fields from only T1 MR images and requires low computation time, suggesting the possibility of using machine learning for realtime electric field estimation in navigated TMS.

Rapid whole-brain electric field mapping in transcranial magnetic stimulation using deep learning

“…Hence, it limits its application in situations when rapid adjustments for multiple coil positions and orientations are needed. To overcome the limitations, several computation algorithms [20][21][22] have been developed to reduce the simulation time.…”

“…The deep-learning technique has been recently applied to predict the E-field induced by TMS [21]. This approach is able to significantly reduce the simulation time to much shorter than one second, a significant reduction in prediction time.…”

“…This approach is able to significantly reduce the simulation time to much shorter than one second, a significant reduction in prediction time. However, the original framework in [21] has several limitations which compromise the accuracy of the estimated E-field. First, it estimates the magnitude of the E-field without any information about the underlying directions which are useful to investigate the effect of TMS on axonal fiber bundles [23].…”

“…First, it estimates the magnitude of the E-field without any information about the underlying directions which are useful to investigate the effect of TMS on axonal fiber bundles [23]. Second, the approach in [21] only predicts E-field with the coil placed near a small brain region around the motor cortex with no training and testing data examined for other brain regions. Third, the neural network takes a T 1w MRI and the position the TMS coil as input to predict the Efield and neglect the difference between coils.…”

Particle Swarm Optimization for Positioning the Coil of Transcranial Magnetic Stimulation

“…Besides PSO, other methods based on brain Atlas [7] and deep neural networks [44] have also shown promise in facilitating accurate brain stimulation.…”