adam2392 / motor-decoding Goto Github PK

Dockerfile 0.01% Makefile 0.01% Jupyter Notebook 99.94% Python 0.06% Shell 0.01%

motor-decoding's Introduction

Analysis of Neural Data Using RF Variants

Here, we use RF, SPORF, SRerf, and MTMORF to analyze the neural data of subjects in a motor task.

Installation

We require the following packages to run experiments:

numpy
scipy
scikit-learn
mlxtend
pingouin
mne
mne-bids
pybv
joblib
tqdm
matplotlib
seaborn
rerf # built locally

Use pipenv with a virtual environment.

# create venv here if not alreayd made
python3.8 -m venv .venv

# install dev packages
pipenv install --dev

# if dev versions are needed
pipenv install https://api.github.com/repos/mne-tools/mne-bids/zipball/master

# make packedForest
cd ../packedForest
make

# pip install into venv
cd ../Python
pipenv run pip install -e .

Note on Compiling PackedForest on Macs

It is recommended to use a conda environment and then follow the instructions on the installation instructions.

motor-decoding's People

Contributors

Watchers

motor-decoding's Issues

Implement class wrapper for SEEG data I/O

Python class that allows easy querying of SEEG data based user-specified events and label keywords.

For instantiation, ideally we only need to specify the subject id
Have functions that allow querying for data - e.g. it returns Epochs data when the user specifies Left Center as an event and target_direction as a label keyword.

Extend data loading functions to include extra trials from the WAR and/or Move task

Extend functions in io/read.py to allow inclusion of fixation data from the WAR task

Plot Trajectories

add code for Efri data
- plot trajectories per trial based on start/stop event key

to accomplish, must:

convert XY data to per trial; get a list of the xy data, where lenght of the list is equal to number of epochs
plot trajectories over time on a fixed 2D graph

Refrence matlab code: https://livejohnshopkins-my.sharepoint.com/:f:/g/personal/ali39_jh_edu/EpJcNaCpsKNNrK9qQrV16bwBW0KHjP45C67KjjyWbC-4hA?e=9fH80D

Write script for decoding movement experiment

Implement Python script that runs the decoding movement experiment and generates summative plots.

Time domain
Frequency domain

Write script for decoding directionality experiment

Implement script that runs the decoding directionality experiment and generates summative plots.

Write script for speed instruction experiment

Implement Python script for speed instruction experiment.

Determine any correlations between decoding performance and speed instruction
Determine any correlations between decoding performance and actual speed

Write script for perceived motion experiment

Implement Python script for visual perceived motion experiment.

Time lock to the "Left Center" event
Use data between "Left Center" and "Hit Target"

Scope out MT-MORF feature importances for R code

Perform quick feasibility test with the feature importances on MNIST data in R.
If loading SEEG data into R is easy, then run MT-MORF feature importances on SEEG data
Scope out how to adapt cpp-R code to cpp-Python

Write script for movement planning experiment

Implement Python script for movement planning experiment.

Time lock to the "Go cue" event and only include data up to "Left Center"

Scope out/write pseudocode for network-based sampling

SUMMARY

Here, we scope out the code to implement any prior biases in multi-variate data. Take X = (X_1, X_2, X_3, ...), where
each X_i has dimensionality, d_i. Then we might have apriori notions of how "correlated" samples are in each axis.

For example:

if X is an image, then this corresponds to just a contiguous patch.
if X is a multivariate time-series, then this corresponds to a discontiguous patch described in neurodata/SPORF#353
if we have further intuition of how to "sample" nearby points that are highly correlated and forming that patch at each node of the decision tree, then we might introduce the notion of a "sampling graph"

TODO

Look into SPORF source code:

Sketch out where new methods would need to be added
Sketch out what these new methods would entail