NIH Blueprint: The Human Connectome Project

Restricted Data Usage

Acceptance of HCP Open Access Data Use Terms is required for access to all HCP data, and will grant the user access to HCP imaging and most behavioral data. Users will be prompted to accept the Data Use Terms upon initial login to ConnectomeDB.

Some data elements, including family structure (e.g. whether a participant is a twin or a non-twin sibling), exact age, handedness, ethnicity and race (see Restricted Data application form for the full list of elements), are available only to qualified investigators who agree to HCP’s Restricted Data Use Terms.  Applications for access to Restricted Data must be submitted by every investigator who will view and use the data (i.e. each member of a research group using the data must apply, from PI to trainees), will be processed individually, and approval is on a case-by-case basis.

An investigator who uses Restricted Data takes on a serious responsibility to protect the privacy of HCP subjects. Specifically, if you are given access to these data, you MUST abide by a prohibition against publishing certain types of individual data  in combination that could make a person individually recognizable, or that could harm and embarrass someone who was inadvertently identified. The specifics of how Restricted Data can be published are detailed in the Restricted Data Use Terms and application (Updated Jan 26, 2016), which qualified investigators must fill out, sign and submit to the HCP.

To provide guidance to investigators, we have developed the following examples to illustrate types of information that can and cannot be published under the conditions that are delineated in the Restricted Data Use Terms.  These examples also illustrate how our data use policy aims to protect subject identity without substantially impeding scientific communication.  If you have either general or specific questions, or if you have additional use case scenarios that would be instructive to consider, please communicate these to Dr. Sandra Curtiss (scurtiss@brainvis.wustl.edu, HCP Project Manager), who will respond directly and potentially add appropriate use cases to this website.

Example 1. Heritability of cortical folding patterns.  

Dr. X, an investigator who qualifies for restricted HCP data access, compares cortical folding patterns in HCP identical twins, non-identical twins, and non-twin sibs.  She discovers that folding patterns are more heritable in some regions than others.  To make her case, she illustrates the findings using two figures, with Figure 1 showing group-average analyses and data, and Figure 2 showing folding patterns for individual subjects for exemplar identical twin pairs vs non-sib pairs.  Figure 2 would necessarily include information about family structure. Under the Restricted Data Use Terms, Dr. X:

  • Could publish Figure 2 showing cortical folding patterns and family structure of selected individuals;
  • Could not report the handedness, exact age, race/ethnicity, body weight, or other restricted access elements of those individuals.
  • Would identify the individuals as Subjects A, B, C, D, E and F in Figure 2.
  • Would not report any HCP subject IDs, either for these individuals or for the subjects included in the group average data, in either the figures or the Methods section of her paper.
  • Would post the HCP subject IDs for Subjects A, B, C, D, E and F and for the subjects included in group averages on the HCP Restricted Access website (through the submit subject key site), so that other qualified investigators could evaluate her results.
  • Could include handedness, exact age, race/ethnicity, or body weight information in Figure 1, as long as the group averages were computed from at least three individuals.

Example 2. Exact Age. 

Qualified Investigator Dr. Y analyzes working memory responses (Z-statistics within a prefrontal ROI) in Task-fMRI scans as a function of age (by year).  He finds a significant correlation and publishes a scatter-plot figure showing this correlation for the ~70 subjects from the HCP Q1 dataset.  He is allowed to publish this scatter plot because, in the plot, age cannot be linked to individual subjects even though the values are plotted as Z-statistics by year for each individual.   In another figure, Dr. Y shows individual-subject surface maps of working memory responses in four of these subjects.   He may not report the age of individual subjects in the figure legend, or even the prefrontal Z-stat score for individual subjects in the legend (because this information would enable someone to look at the working memory scores of the individual subjects and infer, with high probability, their age by year). Under the Restricted Data Use Terms, Dr. Y:

  • Could publish the desired figures, including the HCP subject IDs since no family structure information is provided.
  • Could not report the twin/non-twin status or ethnicity/race of individual subjects in this paper.