Skip to content

Human Whole-Brain Diffusion MRI Dataset, Created in Vivo at Extremely High Resolution, Available Online

Key findings

  • Researchers at Massachusetts General Hospital have created the first publicly available in vivo whole-brain diffusion MRI reference dataset acquired at submillimeter resolution (760 microns)
  • Data were acquired from a single healthy participant across nine 2-hour sessions using the MGH–USC 3-T Connectom MRI scanner equipped with high-strength gradient and a custom-built 64-channel phased-array coil
  • To enable a high signal-to-noise ratio at this extreme resolution, state-of-the-art acquisition techniques were also used along with a carefully designed diffusion protocol
  • The preprocessed diffusion MRI data, T1- and T2-weighted images and field maps for the nine sessions are available for download from the Dryad Digital Repository

The MGH–USC Connectome 3-T MRI scanner housed at Massachusetts General Hospital is a major hardware innovation of the Human Connectome Project, which is attempting to understand the complete details of neural connectivity. The Connectom scanner can produce a magnetic field gradient of up to 300 mT/m strength, which enables in vivo diffusion MRI (dMRI) at extremely detailed resolution.

Researchers Fuyixue Wang, PhD, and Kawin Setsompop, PhD, of the Martinos Center for Biomedical Imaging at Mass General, and colleagues have created the first publicly available in vivo whole-brain dMRI reference dataset acquired at submillimeter resolution—760 microns, a voxel size 7.7 times smaller than that used for conventional dMRI acquisition. They discuss their methods in detail in Scientific Data.

Data Acquisition

The data were acquired in nine 2-hour scanning sessions of a single participant, a healthy white man born in 1989. Besides the Connectom scanner, the researchers made use of:

  • A custom-built 64-channel phased-array coil, as described in Magnetic Resonance in Medicine
  • A custom-made head stabilizer that precisely fit the shape of the participant's head and the inside of the coil, designed to minimize motion contamination and ensure consistent participant positioning across sessions
  • gSlider-SMS, an imaging technique developed at Mass General that allows whole-brain submillimeter acquisition at a high signal-to-noise ratio with low reconstruction errors and limited distortion and blurring, as reported in Magnetic Resonance in Medicine
  • Dual-polarity GRAPPA (generalized autocalibrating partially parallel acquisitions), another method developed at Mass General and reported separately in Magnetic Resonance in Medicine, which integrates ghost correction into the parallel imaging data recovery process

Data Availability and Applications

The preprocessed dMRI data, T1- and T2-weighted images and field maps for the nine sessions are available in compressed NIfTI (Neuroimaging Informatics Technology Initiative) format in the Dryad Digital Repository, here and here. The dataset is expected to have a broad range of uses, including:

  • More detailed examination of the brain's fine-scale structures, such as the gray–white matter boundary and the cellular architecture of thin cortical layers
  • Evaluation of structural brain connectivity
  • Education of medical professionals
  • Testing of new strategies for in vivo high-resolution dMRI on a common platform

Learn more about the Martinos Center for Biomedical Imaging

Learn more about the Department of Radiology


Jyrki Ahveninen, PhD, of the Martinos Center for Biomedical Imaging, and colleagues recently reviewed the use of movies and stories as stimuli during neuroimaging research into memory, attention, language, emotions and social cognition.


An artificial intelligence tool developed by Manisha Bahl, MD, and colleagues identifies patients with low-risk early-stage breast cancer who are appropriate candidates for active surveillance programs, thus helping the patients avoid unnecessary surgery or radiation.