Neuroprint is a software pipeline/Flywheel gear for calculating statistical deviation in cortical thickness from expected values for a healthy individual of the same age and sex. These expected values are derived from a dataset of healthy controls (n=868) spanning the adult age range which were collected from various centers at Penn. The highest-quality T1w image was processed with the ANTs Cortical Thickness pipeline to calculate cortical thickness values in 200 cortical regions delineated by the Schaefer 2018 atlas. A linear regression was performed for each region based on that region’s mean cortical thickness value for the 868 subjects, with age and sex as covariates.
For a given patient, the gear takes the same regional cortical thickness values and the patient’s age and sex and calculates a w-score (essentially a weighted z-score) using the following formula. Note that the score is multiplied by -1 so that higher w-scores reflect greater degeneration.
w-score =-(raw ct val - intercept - age*agecoefficient - sex*sexcoefficient) / standard error residuals
The result is a score for each region describing the difference in that region’s observed cortical thickness compared to that of a healthy control. The gear renders these scores in a 3D model.
While this was originally built for Flywheel, it may be used as a standalone Docker or Singularity image. Here is an example docker run command:
docker run -v /home/will/Gears/neuroprint-pipeline/docker/input:/input \
-v /home/will/Gears/neuroprint-pipeline/docker/output:/output \
willtack/neuroprint-pipeline:0.5.0 \
--label_image_file /input/sub-120870_ses-120870x20191205x3T_Schaefer2018_200Parcels17Networks.nii.gz \
--ct_image_file /input/sub-120870_ses-120870x20191205x3T_CorticalThickness.nii.gz \
--t1_image_file /input/sub-120870_ses-120870x20191205x3T_ExtractedBrain0N4.nii.gz \
--patient_age 72 --patient_sex 0 --thresholds '0.0 0.5 1.0' --prefix sub-120870_ses-120870x20191205x3T \
--output_dir /output
Please see the docker directory for more details.