Scans Acquired by NKI RS1 Study

Cross-sectional and longitudinal lifespan connectomics

Real-Time Neurofeedback

Notes about Multi-Band Imaging

Recently developed multiband echo planar Imaging (Moeller et al., 2010) and multiplexed echo planar imaging (Feinberg et al., 2010) approaches enable the acquisition of functional MRI and diffusion imaging data with unprecedented sampling rates for full-brain coverage through the acquisition of multiple slices simultaneously in the same time it takes to obtain a single slice image using standard EPI (Feinberg et al. (in press), see Smith et al. (2012) for initial application of Multiband EPI with recent improvements (Xu et al. 2012). The Center for Magnetic Resonance Research (CMRR), University of Minnesota, where these sequences are developed for the Washington University-University of Minnesota (WU-Minn) consortium of the Human Connectome Project, has provided the NKI-RS effort with the latest version of the Multiband EPI sequence (Xu et al. 2012) and associated image reconstruction algorithms, enabling the acquisition of state-of-the-art imaging datasets for this large-scale imaging effort. Specific parameter selections were based on initial pilot data to optimize image quality on our scanner. As specified in detail below, three R-fMRI sequences are included in the protocol – one with a TR = 645msec (3mm isotropic voxels, 10 minutes) to provide optimal temporal resolution and another with TR = 1400msec (2mm isotropic voxels, 10 minutes) to provide optimal spatial resolution. A standard EPI sequence (TR = 2500msec, 3mm isotropic voxels, 5 minutes) is included also for reference. We are very grateful to WU-Minn HCP members at CMRR for guidance and assistance in the design, implementation and troubleshooting of these sequences and their reconstruction. These sequences and the associated image reconstruction algorithms for the Siemens platforms are available for distribution from CMRR.

Important Analytic Considerations for Multiband Imaging

Innovations associated with multiband imaging (e.g., simultaneous acquisition of multiple slices) can introduce novel challenges for functional MRI imaging preprocessing and analyses. When using this data please keep in mind the following caveats provided by Christian Beckmann and Steve Smith from the HCP:

  • The increased number of time points obtained when using short TR will change the temporal degrees of freedom. This can be used to drive investigations into the temporal characteristics of resting-state patterns (see Smith et al. (2012)). However, the effective temporal autocorrelation will change. This needs to be corrected for, e.g. when converting regression coefficients or correlation coefficients to T, Z or P statistics (for more discussion see Feinberg et al., 2010)
  • The spatial smoothness of the data may be affected. In particular, smoothness can be non-stationary, so Gaussian-Random Field Theory-based inference (e.g. cluster-correction in in SPM or FSL) might not be suitable.
  • Because multiple slices are excited simultaneously, simple slice time correction will not work and each slice will need to be corrected using custom timing information. Given the short effective TR it is probably less important, though.