Introduction to Gradients¶
Classically, many neuroimaging studies have attempted to parcellate the human brain into distinct areas based on anatomical or functional features. Whilst effective, the strict boundaries assumed by these methods are often unrealistic and there lacks a clear ordering between parcels. More recently, “gradient” approaches, which define the brain as a set of continuous scores along manifold axes, have gained popularity (see also References). Core to these techniques is the computation of an affinity matrix that captures inter-area similarity of a given feature followed by the application of dimensionality reduction techniques to identify a gradual ordering of the input matrix in a lower dimensional manifold space.
BrainSpace is a compact and flexible toolbox that implements a wide variety of approaches to build macroscale gradients from neuroimaging and connectome data. In this overview, we will show the the basic steps needed for the identi cation of gradients, and their visualization. The steps below will help you to get started and to build your first gradients. If you haven’t done so yet, we recommend you install the package (Installation Guide) so you can follow along with the examples.
Basic Gradient Construction¶
The packages comes with the conte69 surface, and several cortical features and parcellations. Let’s start by loading the conte69 surfaces:
To load your own surfaces, you can use the
BrainSpace also provides surface plotting functionality. We can plot the
conte69 surfaces as follows:
Let’s also load the mean connectivity matrix built from a subset of the human connectome project (HCP). The package comes with several example matrices, downsampled using the Schaefer parcellations (Schaefer et al., 2017). Let’s load one of them.
To compute the gradients of our connectivity matrix m we create the GradientMaps object and fit the model to our data:
Now we can visually inspect the gradients. Let’s plot the first gradient:
As we can see, this gradient corresponds to those observed previously in the literature i.e. running from default mode to sensory areas.