Skip to main content
Magnetic resonance imaging datasets with anatomical fiducials for quality control and registration
Alaa Taha, Greydon Gilmore, Mohamad Abbas, Jason Kai, Tristan Kuehn, John Demarco, Geetika Gupta, Daniel Cao, Chris Zajner, Abrar Ahmed, Ali Hadi, Patrick Park, Dimuthu Hemachandra, Reid Vassallo, Sandy Wong, Ali R. Khan, Jonathan C. Lau
Presenting author:
Alaa Taha
Open-source tools available for reproducible, quantitative assessment of brain correspondence have been limited. We previously validated the anatomical fiducial (AFID) placement protocol for point-based assessment of image registration with millimetric (mm) accuracy. The AFIDs protocol involves the manual placement of 32 carefully selected salient points in the human brain, with an inter-rater localization error of 1-2 mm, even by novice raters. Registration accuracy is then assessed as point-based distances between co-registered images.

In this data release, we share manually performed AFID placements and associated structural magnetic resonance imaging (MRI) scans from four datasets (n = 132 subjects total) and more than 10 commonly used MRI templates across field strengths (1.5, 3, and 7T), representing a total of 16,432 individual point placements. Our data is compliant with the Brain Imaging Data Structure (BIDS) allowing for facile incorporation into modern neuroimaging analysis pipelines. Data is accessible on GitHub (github.com/afids/afids-data). In addition, our AFIDs validator (afids-validator.herokuapp.com) facilitates learning of point placements by way of an interactive online user interface.

The shared data can be used to teach neuroanatomy, aid with disease diagnosis, and surgical targeting as demonstrated in previous studies. Upcoming plans include extending the AFIDs protocol to non-human primates and developing applications for automating placement of AFIDs, as well as assessment of image registration.