A recent study unveiled a new statistical method that was able to analyze human brain white matter to accurately classify patients with amyotrophic lateral sclerosis (ALS) in one dataset and designate “brain age” in another.
White matter contains long-range connections between different brain regions. The organization of these connections reveals information about brain function and health, providing insight into cognitive skills and predicting clinical symptoms across a variety of psychiatric and neurological disorders.
Prior statistical inference approaches from tractometry, a method that uses diffusion-weighted magnetic resonance (dMRI) data to quantify brain tissue properties, have been limited in either sensitivity or statistical power. The new method resolves these issues by using a regularization technique called Sparse Group Lasso that was developed by UW Associate Professor of Biostatistics, Noah Simon.
The study analysis looked at two datasets. In one, the new method was able to accurately classify patients with ALS and detect that ALS is a disease of the motor system. In another dataset., the method accurately designated “brain age” based on the properties of the brain connections.
Authors expect the method will be useful in exploring other questions and datasets and have made their software available open source to other researchers.
Funding support for this research was provided by the National Institute of Mental Health (NIMH) and the USBrainAlliance.