Neural Network Model for Estimation of the Induced Electric Field During Transcranial Magnetic Stimulation

Afuwape, Oluwaponmile F.; Olafasakin, Olumide O.; Jiles, David

doi:10.1109/tmag.2021.3086761

Cited by 13 publications

(18 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, a rather large body of work has explored the power of deep neural networks for computing the electrical activity induced by TMS in the brain (Yarossi et al, 2019;Yokota et al, 2019;Akbar et al, 2020;Afuwape et al, 2021;Sathi et al, 2021). While these works can be differentiated based on the exact specification of input and output variables, most of these models can be better categorized based on the direct or indirect usage of magnetic resonance (MR) images along with the coil parameters.…”

Section: Stimulus-response Modelsmentioning

confidence: 99%

“…While using processed MR images can help simplify the modeling (by requiring smaller network size and lesser data, for example), this step is often computationally expensive and relies on conduction models that are manually tuned. In contrast, MR images and coil parameters have been used directly as inputs to a convolution neural network to determine either the whole-brain electric field distribution (Yokota et al, 2019) or the electric field statistics such as maximum induced electric field (Afuwape et al, 2021). While the models in the direct approach were trained and tested using simulated TMS data, the predictability achieved with these models combined with the computational performance that comes with not having to pre-process MR images still provides a case for end-to-end data-driven modeling.…”

Section: Stimulus-response Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Brain modeling for control: A review

Acharya

Ruf

Nozari

2022

Front. Control Eng.

View full text Add to dashboard Cite

Neurostimulation technologies have seen a recent surge in interest from the neuroscience and controls communities alike due to their proven potential to treat conditions such as epilepsy, Parkinson’s Disease, and depression. The provided stimulation can be of different types, such as electric, magnetic, and optogenetic, and is generally applied to a specific region of the brain in order to drive the local and/or global neural dynamics to a desired state of (in)activity. For most neurostimulation techniques, however, an underlying theoretical understanding of their efficacy is still lacking. From a control-theoretic perspective, it is important to understand how each stimulus modality interacts with the inherent complex network dynamics of the brain in order to assess the controllability of the system and develop neurophysiologically relevant computational models that can be used to design the stimulation profile systematically and in closed loop. In this paper, we review the computational modeling studies of 1) deep brain stimulation, 2) transcranial magnetic stimulation, 3) direct current stimulation, 4) transcranial electrical stimulation, and 5) optogenetics as five of the most popular and commonly used neurostimulation technologies in research and clinical settings. For each technology, we split the reviewed studies into 1) theory-driven biophysical models capturing the low-level physics of the interactions between the stimulation source and neuronal tissue, 2) data-driven stimulus-response models which capture the end-to-end effects of stimulation on various biomarkers of interest, and 3) data-driven dynamical system models that extract the precise dynamics of the brain’s response to neurostimulation from neural data. While our focus is particularly on the latter category due to their greater utility in control design, we review key works in the former two categories as the basis and context in which dynamical system models have been and will be developed. In all cases, we highlight the strength and weaknesses of the reviewed works and conclude the review with discussions on outstanding challenges and critical avenues for future work.

show abstract

Section: Stimulus-response Modelsmentioning

confidence: 99%

Section: Stimulus-response Modelsmentioning

confidence: 99%

Brain modeling for control: A review

Acharya

Ruf

Nozari

2022

Front. Control Eng.

View full text Add to dashboard Cite

show abstract

“…The encoder and decoder-based U-Net model performed electric field prediction with an accuracy of 93%. In another study, Afuwape et al 16 utilized a convolutional neural network (CNN) based DL model to predict electric fields from sixteen different single-type TMS coils. Initially, the creation of the dataset was performed in Sim4Life software by simulating the different coils on the 3D human head that was generated from T1-weighted MRI.…”

Section: Introductionmentioning

confidence: 99%

“…However, the data creation processes from the segmentation software presented in Refs. 15 , 16 are quite challenging because it requires high contrast T1-weighted MRI brain image. It is also difficult to produce an actual human head model from low-contrast MRI images.…”

Section: Introductionmentioning

confidence: 99%

Attention-assisted hybrid 1D CNN-BiLSTM model for predicting electric field induced by transcranial magnetic stimulation coil

Sathi

Hosain

Hossain

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Deep learning-based models such as deep neural network (DNN) and convolutional neural network (CNN) have recently been established as state-of-the-art for enumerating electric fields from transcranial magnetic stimulation coil. One of the main challenges related to this electric field enumeration is the prediction time and accuracy. Despite the low computational cost, the performance of the existing prediction models for electric field enumeration is quite inefficient. This study proposes a 1D CNN-based bi-directional long short-term memory (BiLSTM) model with an attention mechanism to predict electric field induced by a transcranial magnetic stimulation coil. The model employs three consecutive 1D CNN layers followed by the BiLSTM layer for extracting deep features. After that, the weights of the deep features are redistributed and integrated by the attention mechanism and a fully connected layer is utilized for the prediction. For the prediction purpose, six input features including coil turns of single wing, coil thickness, coil diameter, distance between two wings, distance between head and coil position, and angle between two wings of coil are mapped with the output of the electric field. The performance evaluation is conducted based on four verification metrics (e.g. R2, MSE, MAE, and RMSE) between the simulated data and predicted data. The results indicate that the proposed model outperforms existing DNN and CNN models in predicting the induced electrical field with R2 = 0.9992, MSE = 0.0005, MAE = 0.0188, and RMSE = 0.0228 in the testing stage.

show abstract

“…In addition, subjects requiring TMS will not be exposed to unnecessary stimulation as the DCNN model can predict the TMS responses for the specific subject. Other studies [35], [36] slightly modified the DCNN algorithms to overcome a few limitations and achieved a better performance in predicting E-fields compared to previous studies. Previous studies have relied on a public database to obtain MRIs and augment the data by rotating the coil position, which does not represent the clinical conditions as the coil position and rotation do not change significantly compared to the shape and size of the head.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Electric Fields Induced by Transcranial Magnetic Stimulation in the Brain using a Deep Encoder-Decoder Convolutional Neural Network

Tashli

Alam

Gong

et al. 2022

Preprint

View full text Add to dashboard Cite

Transcranial magnetic stimulation (TMS) is a non-invasive, effective, and safe neuromodulation technique to diagnose and treat neurological and psychiatric disorders. However, the complexity and heterogeneity of the brain composition and structure pose a challenge in accurately determining whether critical brain regions have received the right level of induced electric field. Numerical computation methods, like finite element analysis (FEA), can be used to estimate electric field distribution. However, these methods need exceedingly high computational resources and are time-consuming. In this work, we developed a deep convolutional neural network (DCNN) encoder-decoder model to predict induced electric fields, in real-time, from T1-weighted and T2-weighted magnetic resonance imaging (MRI) based anatomical slices. We recruited 11 healthy subjects and applied TMS to the primary motor cortex to measure resting motor thresholds. Head models were developed from MRIs of the subjects using the SimNIBS pipeline. Head model overall size was scaled to 20 new size scales for each subject to form a total of 231 head models. Scaling was done to increase the number of input data representing different head model sizes. Sim4Life, a FEA software, was used to compute the induced electric fields, which served as the DCNN training data. For the trained network, the peak signal to noise ratios of the training and testing data were 32.83dB and 28.01dB, respectively. The key contribution of our model is the ability to predict the induced electric fields in real-time and thereby accurately and efficiently predict the TMS strength needed in targeted brain regions.

show abstract

Neural Network Model for Estimation of the Induced Electric Field During Transcranial Magnetic Stimulation

Cited by 13 publications

References 30 publications

Brain modeling for control: A review

Brain modeling for control: A review

Attention-assisted hybrid 1D CNN-BiLSTM model for predicting electric field induced by transcranial magnetic stimulation coil

Prediction of Electric Fields Induced by Transcranial Magnetic Stimulation in the Brain using a Deep Encoder-Decoder Convolutional Neural Network

Contact Info

Product

Resources

About