Conversation

And brain scanners aren’t in pediatric clinics (yet), but if they were, what would we measure? How “should” the brain look? What are the milestones to reach? Unlike height/weight we don’t have a universal standard to anchor brain metrics…
1
4
But we can use proven statistical frameworks to help stitch datasets together, which is what we did here across >100,000 people (>125,000 brain scans) from over 100 studies. Some scans dated back to 1990!
1
3
Taken together, since each brain tissue matures at different rates, we can map new milestones when the underlying distribution of tissues change most. For example the period right after birth when grey matter grows much faster than white matter (grey/white differentiation).
1
3
Converting brain features into (per)centile scores allows for assessment across metrics, the creation of combined metrics, and comparison of patient groups that are generally diagnosed at different ages. Amazing to compare within and across clinically diverse conditions!
1
2
What’s really cool is that we can leverage our massive combined dataset to provide tools for converting brain tissue quantities into percentile scores for other existing and new cohorts, which we call "out-of-sample" tools, available to use now!
1
4
We also did an extensive set of supplemental analyses to test robustness against manual ratings and automated measures of image quality, other imaging modalities, and a range of additional clinical conditions. Check out some of our ~200 pages of the supplemental information!
1
2