motorTree

The motorTree pipeline builds on the innovative concept of the “lifespan tree of brain anatomy”, transforming raw MRI data into intuitive maps that track disease-related brain changes across the lifespan. By modeling the trajectories of more than one hundred brain structures, it uncovers the subtle anatomical signatures that distinguish between overlapping neurodegenerative conditions. Leveraging advanced dimensionality reduction and smart data augmentation, motorTree generates interpretable and clinically meaningful patterns, moving beyond opaque “black-box” outputs. The result is a powerful diagnostic assistant that not only surpasses conventional AI models but also equips clinicians with clearer insights for patient care.

Specifically, motorTree is trained to discriminate between Parkinson’s disease (PD), Multiple System Atrophy (MSA), Progressive Supranuclear Palsy (PSP), and Dementia with Lewy Bodies (DLB). The lifespan tree model provides a probability score for each motor disorder, reflecting the subject’s proximity to the tree branches representing the four conditions. In addition, motorTree delivers a map of brain structures deviating from normal aging, highlighting the regions that drive the final diagnosis.

All the considered structures are segmented using AssemblyNet

Please note that motorTree is designed exclusively to distinguish PD, PSP, MSA, and DLB. It should not be applied to other pathologies or to patient without any parkinson-like motor symptoms (since there is no control class).

Report

Once the process is finished you will be notified by e-mail so you will be able to download a package including some image files and two (CSV and PDF) reports presenting the probability for the four considered classes (i.e., PD, DLB, MSA, and PSP). Moreover, the report provides an abnormality map indicating the most divergent structures compared to normal aging.