Freesurfer Dataset

This is simulated data created for test and demonstration purposes for decentralized algorithms for COINSTAC.

There is a CSV file with the two columns of demographic data and a column of filenames containing the volumes of the regions of interest (ROIs) of the subjects' brains. These volumes are typically produced by segmentation using recon-all in the FreeSurfer program.

Data dictionary

• freesurferfile: filename associated with subject containing volumes of ROIs, as typically produced by FreeSurfer segmentation
• age (float): subject's age
• isControl (Boolean): whether subject was a control (True) or had a medical condition (False)

There are a total of 72 subjects.

The volumes were simulated with equations of the following form:

y = B0 + B1 x age + B2 x isControl + B3 x e

where e ~ N(0,1)

Each volume was given a fixed intercept (B0) (e.g., 48466.3 for Right-Cerebellum-Cortex). The effect of age (B1) was selected randomly from the range [-300, -100], and the effect of diagnosis (isControl, B2) was selected randomly from the range [500, 1000] for each pseudo subject. A standard unit Gaussian noise multiplied by a random index from the range [1800, 2200] was added subsequently.

This data was created on Feb 2, 2017 by Jing Ming at the Mind Research Network and presented in this paper:

Ming J, Verner E, Sarwate A et al. COINSTAC: Decentralizing the future of brain imaging analysis [version 1; peer review: 2 approved]. F1000Research 2017, 6:1512 (https://doi.org/10.12688/f1000research.12353.1)