Collecting brain scans from thousands of people can be challenging in autism research; data-sharing and collaborative efforts can help drive results that stand up to statistical scrutiny.

The machine-learning approach could help identify how brain structure differs between autistic and non-autistic infants, the researchers say.

Many brain regions develop differently between people with 22q11.2 duplications and deletions, and those trajectories also vary with a person’s diagnosis.

Regions of the brain’s fear center expand in autistic children and teenagers with anxiety, but not in their autistic or non-autistic peers without anxiety.

By as early as age 2, autistic children appear to have a smaller salience network and a larger default mode network, among other differences, than children without the condition.

The new resource is the first to chart human brain development from before birth to 100 years of age.

Animals with different autism-linked mutations share disruptions to the mTOR signaling pathway, pointing to a potential molecular mechanism for the atypical cerebellar development seen in some autistic people.

Having a genetic predisposition to inflammation is linked to structural changes in brain regions implicated in neurodevelopmental conditions.

An overreliance on small studies with limited reproducibility has slowed the advancement of neuroimaging, a new analysis suggests.

Model mice of the subtype also show hyperactivity in a signaling pathway called mTOR, bolstering the idea that distinct forms of autism have different biological roots and may require different treatment approaches.