Motor Cortex Mapping (Forward and Inverse) using TMS-EMG and ConvNets

Project Summary

Overall System:

Modeling Pipelines:

Inverse Imaging System Network Architecture:

File Structure

|--<data>
|--coding file (e.g. M2M-InvNet\train.py)

Usage

An execution sample for training the Direct Variational model from M2M-InvNet:

python train.py --dir <str, data directory> --sub <int, subject number>

An execution sample for testing the Direct Variational model from M2M-InvNet:

python test.py --dir <str, data directory> --sub <int, subject number>

test.py will generate fold-level statistics (NRMSE, R^2) for the selected subject. Use eval.py (currently in a Jupyter cell format) for more advanced inference features. Supporting functions are automatically parsed from utils.py. All other models implemented in our paper can be found in allModels.py.

The models from M2M-Net are in a Jupyter cell format. Please consider running the cells sequentially in VSCode or Spyder.

Matlab scripts used to preprocess and visualize both the forward and inverse models can be found in other.

Publications

Please take a look at our papers below (and cite if you find helpful), for the corresponding coding folders:

[1] M2M-InvNet Inverse Model (EMG->TMS: 2024)

Cite:

@article{akbar2024m2m,
  title={M2M-InvNet: Human Motor Cortex Mapping from Multi-Muscle Response Using TMS and Generative 3D Convolutional Network},
  author={Akbar, Md Navid and Yarossi, Mathew and Rampersad, Sumientra and Lockwood, Kyle and Masoomi, Aria and Tunik, Eugene and Brooks, Dana and Erdo{\u{g}}mu{\c{s}}, Deniz},
  journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
  year={2024},
  publisher={IEEE}
}

[2] M2M-Net Forward Model (TMS->EMG: 2020)

Cite:

@inproceedings{akbar2020m2m,
author = {Akbar, Md Navid and Yarossi, Mathew and Martinez-Gost, Marc and Sommer, Marc A. and Dannhauer, Moritz and Rampersad, Sumientra and Brooks, Dana and Tunik, Eugene and Erdo\u{g}mu\c{s}, Deniz},
title = {Mapping Motor Cortex Stimulation to Muscle Responses: A Deep Neural Network Modeling Approach},
year = {2020},
booktitle = {Proceedings of the 13th ACM International Conference on PErvasive Technologies Related to Assistive Environments},
numpages = {6}
}