A Collection of fMRI Tools
This site provides a landing page to summarize resources within the ME-ICA organization.
Note
ME-ICA was originally created specifically for the tedana
library,
which uses independent components analysis (ICA) on multi-echo (ME) fMRI
data. As part of development, we built tools that work with tedana
but may also have stand-alone uses for processing and visualizing fMRI
data that is not multi-echo. While the "ME-ICA" may not accurately
describe all these tools, we have chosen to keep it for now because we're
used to it.
tedana
tedana TE Dependant ANAlysis. Calculates echo dependance of each ICA component and denoises data by removing components that are unlikely to contain BOLD signal.
rica
rica Reports for ICA. An app to visualize ICA components and perform manual classification in an interactive way. This could work on any component maps of
fMRI data. It currently expects some files that are output from tedana
but similar files could be generated by other programs or addition input
options could be added to rica
mapca
mapca Moving Average Principal Components Analysis. This started as a Python port of the dimensionality reduction technique described in Li et al 2007 and used in GIFT.
ddmra
ddmra is a Python package for performing distance-dependent motion-related artifact analyses. This is modeled after the analyses performed in Power et al 2018
me-ica
meica The package that started it all. This is a clone of the multi-echo denoising code by Prantik Kundu that tedana
is based on. It is no longer maintained, and only runs on Python 2.7. It may be useful as a reference to to replicate past work, but we
recommend people use tedana
instead of this.
Works in progress
We welcome contributors, but these tools are not ready for general use.
Multi-Echo (fMRI) Data Analysis Book
multi-echo-data-analysis An early-stage effort to create a Jupyter-based
book to explain multi-echo fMRI analysis and tedana
. Comments, ideas,
and contributions are very welcome.
aroma
aroma is a fork of ICA-AROMA
which we have been working to make work purely with Python, in a
BIDS-compatible manner. AROMA is similar to tedana
in that it uses
features of ICA components to classify some as noise to remove them. Tedana's
features are primarily based on how signals change across echoes. AROMA's
features are based on how much of a component's highly weighted voxels are
near brain or ventricle edges and whether component time series are too high
frequency to be BOLD. There are complementary approaches that may work well
together.
godec
godec is a Python package for Go Decomposition, specifically built to work with fMRI data.
Scientific Conferences
We have documented the progress of tedana at the Organization For Human Brain Mapping annual meetings. For several years, we have shared the tedana poster, an interactive demonstration of the results and compiled other multi-echo posters and presentations at the conference. Each year includes an interactive demonstration of how tedana's output looked at that time.