Skip to main content
2024 BMES Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Neural Engineering

Neural Engineering - Poster Session D

Poster J20 - Motor-Evoked Potential Response to Transcranial Magnetic Stimulation for Development of a Deep Learning Model

Friday, October 25, 2024
3:30 PM - 4:30 PM EST
Location: Exhibit Hall E, F & G

Introduction:  Repetitive transcranial magnetic stimulation (TMS) is able to noninvasively modulate a specific functional region of the brain, either directly or indirectly, by targeting accessible regions of a specific network. Although TMS is widely and successfully used for the treatment of major depressive disorder and for brain mapping, stimulation parameters have not been systematically evaluated. Without standardization of stimulation protocols, outcomes have been inconsistent, hindering its more widespread clinical application. TMS parameters used for treatment or diagnosis depend on the brain neuroanatomy [1] and functional connectivity of the subjects. Functional connectivity among the brain regions is highly varied between subjects, yet, its effects on TMS have not been studied well. Therefore, it will be important to establish whether jointly accounting for individual neuroanatomy and functional connectivity can lead to standardized TMS stimulation protocols and improved clinical outcomes. Here, we present the TMS response through motor-evoked potentials (MEPs) that will be used in conjunction with neuroanatomy and functional connectivity to train a deep convolutional neural network to predict TMS response.

Materials and

Methods:  Electromyography (EMG) data were recorded using a Trigno™ Wireless System (Delsys,MA). EMG signals were recorded with Spike 2 software (Cambridge Electron Design, Cambridge, UK). The FDI and first palmer interosseus (FPI) were instrumented with surface electrodes on the skin. EMG signals will be amplified (x1000), bandpass filtered (20-450Hz) prior to A/D conversion (Micro 1401, Cambridge Electron Design, UK), and sampled at 2kHz. Single pulse TMS was delivered as a monophasic current to the primary motor cortex contralateral to the resting arm using a Magstim BiStim2 stimulator via a 70 mm figure-of-eight coil. The vertex at the intersection of the inion-nasion and inter-aural lines are marked on a cap fitted on the participant’s head and used to orient the coil near the cortical target. The coil was held tangentially on the scalp via a support stand with the coil center rotated to induce a posterior-to-anterior cortical current across the central sulcus. The hotspot for the FDI was identified as the location evoking the largest peak-to-peak amplitude MEP using the lowest stimulation intensity [1]. Resting motor threshold was determined [1] and TMS was administered using computerized control, at a range of stimulus intensities from 80%-160% of participants’ respective RMT in random order, featuring 5 second inter- stimulus intervals. Twenty MEPs were recorded at each stimulus intensity, such that 120 MEPs in total were recorded per participant.

Results, Conclusions, and Discussions: Preliminary data from healthy subjects demonstrate preferential targeting of the FDI muscle with TMS, relative to the FPI muscle (Figure 1). Average RMT values across subjects was 43% MSO. As the stimulation intensity increases above RMT, MEPs increase (Figure 1). The Boltzmann function fit the TMS response data better than a linear fit. The slope of the MEP input-output curve indicates the excitability of the targeted neuromuscular circuit across multiple neuronal populations. MEPs can be measured after stimulation by TMS of the relevant motor area of the brain (FDI motor hotspot) and is maximized for a given TMS signal by optimizing the region where the TMS is applied. However, the MEP magnitude will also be influenced by anatomical, connectivity and functional details of the brain that are characterized by magnetic resonance imaging (MRI), diffusion tensor imaging (DTI), and functional MRI images [3]. Our on-going work aims to deliver a process to determine TMS response in the form of MEPs in human subjects using deep convolutional neural networks, informed by MRI, DTI, and fMRI. The outcome of our project will be a method to predict motor thresholds needed to determine the TMS treatment dosage for treating various neurological and psychiatric disorders. The successful outcome of this work will also allow researchers to minimize the new data that needs to be generated for training DCNNs by using transfer learning of related publicly available data.

Acknowledgements (Optional):  Acknowledgements: This work is supported by the National Science Foundation Research Experience for Undergraduate in Magnetics at VCU (DMR-2349694), and the VCU Breakthroughs Fund.

References: [1] Mittal, N., et al. Human Brain Mapping 2022, 43(14):4492-4507. [2] P. M. Rossini et al., Clin Neurophysiol Suppl, vol. 52, pp. 171-85, 1999. [3] A. Zandvakili, et al. Eur. Arch. Psychiatry Clin. Neurosci., vol. 271, no. 1, pp. 29–37, 2021.