Tractometry

Tractometry uses diffusion-weighted MRI (dMRI) to extract microstructural tissue properties of major white matter pathways. Here, we maintain a suite of integrated, open-source software that performs all analysis stages:

pyAFQ: Automated Fiber Quantification in Python:
 • post-processing of dMRI data
 • delineation of major white matter pathways
 • modeling of the tissue properties within them
 • expects pre-processed data. Data can be pre-processed with QSIprep

AFQ-Insight: Machine learning and statistics for tractomtery:
 • novel machine learning models such as convolutional neural networks (CNNs) and recurrent neural network (RNNs)
 • more standard approaches, such as ordinary least squares (OLS) and principal component analysis (PCA)
 • integrated with scikit-learn and adapted for tract data, bridging these two worlds

Tractable: R-based statistical analysis of tractometry:
 • focuses on generalized additive models (GAMs)

AFQ-Browser: Interactive exploratory visualization and sharing of tractometry studies:
 • allows researchers to interactively query the data to explore patterns

Tractoscope: Visualization of large openly-available tractometry studies.

Examples

pyAFQ

• Basics

  • Introduction
  • Visualizing AFQ derivatives

• Extensions

  • BabyAFQ: tractometry for infant dMRI data
  • RecoBundles for tract delineation

• Adding New Bundles

  • Optic Radiations
  • Acoustic Radiations
  • SLF 1/2/3 Subdivisions

• Acceleration

  • GPU Tractography
  • Multiprocessing for Model Fitting (Ray)

AFQ-Insight

• Parametric Statistics for Group Comparison

• Regression: Predict Age from White Matter

• Classification: ALS Diagnosis

Tractable

• Fitting Generalized Additive Models (GAMs)

How to get help

We encourage you to seek help and share your questions with the community. Here’s how to get support for Tractometry-related projects:

1. Check NeuroStars
   NeuroStars is a community forum for neuroimaging questions. Search for existing answers or post your question using the pyafq, afq-insight, or other relevant tags.

2. Browse or open issues on the respective GitHub repositories
   Many questions may already be answered in the project’s issue tracker. If not, you can open a new issue:

   • pyAFQ

   • AFQ-Insight

   • Tractable

   • AFQ-Browser

   • Tractoscope

3. Include details when asking for help
   When posting, please include:

   • The software version you’re using

   • Relevant code or command-line calls

   • Error messages (if any)

   • Expected vs. actual behavior