Multi-subject, multi-modal (fMRI+MEG+EEG) neuroimaging dataset on face recognition ================================================================================== These data are provided freely for research purposes only (as part of their Award of the BioMag2010 Data Competition). Note: total data size is ~120GB (and you will need at least 20 GB extra free space for derived data if you use SPM8 scripts). There are 16 subjects, and within each subject directory ("Sub%02d"), you can find: 1. Structural MRI image in "T1" subdirectory, containing: 1.1 Single NIFTI MPRAGE T1-weighted image ("mprage.nii") from a Siemens 3T Trio. !!Note that the face has been deleted from these images to protect identity!! 1.2 Text file of manually-determined 3D locations of fiducials in space of above image ("mri_fids.txt"): a 3 x 4 table with x,y,z location (mm) and fiducial name (NAS = nasion; LPA = left pre-auricular; RPA = right pre-auricular) 1.3 Matlab version of above x,y,z locations of fiducials ("mri_fids.mat") 1.4 Text file containing digitised headpoints (in x,y,z mm relative to space defined by fiducials) ("headpoints.txt") !! Note that digitised points on below the forehead (eg nose) have been removed because T1s de-faced (above), and because nose is not reproduced fully on T1 anyway!! 2. Functional MRI data in "BOLD" subdirectory, containing: 2.1 Further subdirectories for the 9 runs (sessions) ("Run_%02d"), each containing 208 "fM*.nii" fMRI BOLD EPI volumes (33 slices, 3x3x3.75mm voxels, TR=2s, ~7mins), recorded within one visit (2 dummy scans already removed). !!Note that, owing to scanner error, Subject 10 only has 170 volumes in last run (Run 9)!! 2.2 A subdirectory called "Trials" that contains a text and matlab file for each run, which code the onsets of each trial of each condition in SPM format (ie, onsets in units of seconds, condition names, and durations (always 0 here to treat as events)). Note that periods of fixation between blocks of intermixed events were inserted (unlike for MEEG version below), to provide a baseline. 2.3 A subdirectory called "FieldMap" that contains 3 images - 2 magnitude images at 2 echo-times (5.19ms and 7.65ms) and resulting 1 phase difference image - which can be used for unwarping the BOLD EPI images above. Note that the MRI (and fMRI) data have been converted from original Siemens DICOM format to NIFTI (via spm_dicom_convert); for original raw files, contact person below. 3. MEG and conccurrent EEG data in a "MEEG" subdirectory, containing two files per run: 3.1 A raw Neuromag FIF file (*raw*.fif) of continuous data for each of 6, ~9mins runs ("run_%02d_sss.fif") at 1.1Khz sampling from 102 magnetometers, 204 planar gradiometers and 70 EEG electrodes in a Elekta VectorView system, recorded on a second visit of each subject. 3.2 A MaxFiltered version of each continuous file (*sss*.fif), which has been de-noised and movement-corrected using the proprietary MaxFilter v2.2 software that implements (temporal) Signal Space Separation. 3.3 A subdirectory called "MaxFilterOuput" that contains text files describing the output of the above MaxFilter step. 3.4 A subdirectory called "Trials" that contains a text and matlab file for each run, which code the type of each trial, and the time from start of run of a 1700ms epoch (including its start time, stop time and prestimulus period (-500ms, or 550 samples), in FieldTrip(SPM8) "trl" format). 4. Diffusion-Weighted Images (e.g, for DTI) in a "DWI" subdirectory, containing though ONLY for 11 of the 16 subjects (1, 2, 3, 4, 5, 6, 9, 12, 13, 14, 15): 4.1 A 4D NIFTI image ("dwi.nii") 4.2 A text file containing the 65 B values ("dwi.bval") 4.3 A text file containing the 65 gradient directions ("dwi.bvec") 5. Structural MRI image in "FLASH" subdirectory, containing: 5.1 14 *.mgz files (see http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/MghFormat) from two bandwidth-matched 1x1x1 multi-echo FLASH sequences (giving better skull contrast than "T1" image above), with 7 echo-times (1:85ms; 4:15ms; 6:45ms; 8:75ms; 11:05ms; 13:35ms; 15:65ms; labelled "0"-"6" in files) at each of 2 flip-angles (5 and 30 degrees). There are also another top-level directory: A "EmptyRoom" directory, which contains 3 sub-directories of empty-room recordings of 3-5mins acquired at roughly the same time of year (spring 2009) as the 16 subjects. The sub-directory names are Year (first 2 digits), Month (second 2 digits) and Day (third 2 digits). Inside each are 2 raw *.fif files: one for which basic SSS has been applied by maxfilter in a similar manner to the subject data above, and one (*-noSSS.fif) for which SSS has not been applied (though the data have been passed through maxfilter just to convert to float format). The *.log files for the maxfilter run are there too, as are versions that have been converted into SPM8 (the *.mat and *.dat files). Further empty-room recordings, acquired closer to the recordings from each subject, can be obtained from rik.henson@mrc-cbu.cam.ac.uk on request. A "Publications" directory containing some PDFs that describe parardigm, data and results in more detail (further publications resulting from these data can be deposited here too if desired). A "Stimuli" directory containing *.bmp BitMap images all of the stimuli used, where initial "f" in filename refers to famous, "u" refers to unfamiliar and "s" refers to scrambled. Note that only 3 conditions (event-types) are included in the Trials directories: famous faces, unfamiliar faces and (phase)-scrambled faces. The full experiment included a further 3-level factor of first presentation, immediate repetition and delayed repetition, which has been collapsed here (ie data will be averaged across initial and repeated presentations). The "fame" of faces is also based on the stimuli: in the full experiment, there was a debriefing session in which participants indicated whether they knew each face, but these behavioural data have not been used to adjust conditions here (recognition performance was fairly accurate). If you want to use SPM8 to analyse these data, as in the Wakeman & Henson (in prep) paper, you can obtain analysis scripts from ftp://ftp.mrc-cbu.cam.ac.uk/personal/rik.henson/SPMScripts Data from another 3-5 subjects, but who did not complete all parts of the experiment, or had technical problems, are available on request. EPI fieldmaps and (1x1x8) T2 images are also available for each subject on request. If you wish to publish any of these data, please acknowledge Daniel Wakeman and Richard Henson. The data has been published so far in (see "Publications" sub-directory for PDFs): Wakeman, D.G. & Henson, R.N. (2010). Functional and structural connectivity in face-processing: MEG, EEG, fMRI, MRI and DWI data. BioMag 2010 Conference. (This is the Poster associated with the BioMag 2010 Data Competition Prize that these data were awarded). Wakeman, D.G. & Henson, R.N. (in prep). A multi-subject, multi-modal human neuroimaging dataset. (This will be the best citation to describe the basic paradigm and data). Henson, R.N., Wakeman, D.G., Litvak, V. & Friston, K.J. (2011). A Parametric Empirical Bayesian framework for the EEG/MEG inverse problem: generative models for multisubject and multimodal integration. Frontiers in Human Neuroscience, 5, 76, 1-16. Henson, R.N., Wakeman, D.G., Phillips, C. & Rowe, J. (2012). Effective Connectivity between OFA and FFA during face perception: DCM of evoked MEG, EEG and fMRI responses. HBM2012 Abstract. Any questions, please contact rik.henson@mrc-cbu.cam.ac.uk. Created: 27-07-2012 Updated: 19-09-2012 (DWI data added) Updated: 25-01-2013 (raw FIF files added, and SPM scripts moved to another FTP directory) Updated: 22-02-2013 (FTP transfer errors on some files corrected - thanks to Justin Chumbley Updated: 20-03-2013 (Empty Room data - thanks to suggestion of JC Chung) Updated: 20-12-2013 (FTP-error for BOLD for one subject - thanks to Ged Ridgway) Updated: 06-01-2013 (FLASH data for each subject - thanks to Dan Wakeman) Updated: 14-03-2014 (FieldMap EPI data for each subject)