Overview of capabilities

A complete streamline generation workflow typically involves the following steps:

Estimating FODs
Choosing a model for the streamlines
Choosing a model for estimating fibre orientation distributions (FODs)
Generating seeds from the FODs
Running tractography algorithms to produce streamlines

The Tractography.jl package focuses on steps >1. Step 1 — estimating FODs — should be performed using a dedicated library, such as Fibers.jl.

Model of streamlines specification

The model of strealimes can either be

Tractography Markov Chain
Diffusion
Transport

In practice, the general model is specified by a Model while the simulation algorithm Probabilistic, Deterministic, Diffusion, Transport selects the model of streamlines.

FOD specification

The FOD can be supplied to the model by the keyword argument foddata. See Tractography.Model for more details. Now, you can pass to foddata a Tractography.FODData object (see Tractography.FODData for more information), which can be created:

from a NIfTI file using the simplified Tractography.FODData constructor.
from an AbstractArray object using directly the Tractography.FODData constructor.

Seed Generation

Seeds can be generated directly from the FOD or from a binary mask using the following methods:

Tracking Algorithms

Four tracking algorithms are currently implemented (see Algorithms).
All algorithms can run on either GPU or CPU (threaded), and are vendor-agnostic.
The interface supports precomputed caches, making the algorithms allocation-free during execution. This allows for efficient, long-duration runs.
Computations support arbitrary floating-point precision (Float32, Float64, BigFloat, ...).

Plotting

The package provides convenient plotting utilities to:

Visualize streamlines
Display FODs using glyph-based representations

Exporting Streamlines

Generated streamlines can be exported in the .tck format. Alternatively, users can export data to any preferred format using standard Julia packages (e.g. JLD2.jl).