Data Issues

fMRI TR stated incorrectly in dataset

Converted fMRI data contains incorrect TR of 1 second in NIfTI headers when in fact the TR was 2 seconds. This was due to a bug in the software used to convert from DCM to NIfTI. The issue has been corrected in the latest version of the BIDS dataset (see JSON sidecar file for correct TR).

BIDS Dataset Released

MRI and MEG data have been updated to conform with latest BIDS standards. MRI data has been converted using the dcm2bids package, while MEG data has been converted using the fieldtrip bids converter.

MEG Head Position Indicators

Note first 30-40s of each MEG run was without continuous Head Position Indicator (cHPI) coils switched on (note this does not affect trial onsets, which always started after cHPI turned on, but does affect resting-state).

MRI Coil Indicator

In October 2011, after the first 97 CamCAN participants had been acquired, the 3T TRIO was ramped down after the gradient coil failed, and therefore B0 frequency and passive shimming configuration differed for the subsequent participants. The coil-type is now available as a binary variable within the standard tabular data folder provided for all requests.

fMRI SMT Subjects Removed

Due to a problem with fMRI data acquisition, two subjects were removed from the fMRI SMT dataset (IDs CC610462 and CC220519).

MEG Subjects Removed

Due to a problem with HPI coils, two subjects were removed from the MEG dataset. For those who already have downloaded data before the above date, IDs CC210314 and CC221733 should be removed from the dataset.

DWI Subjects Removed

After quality control of the diffusion weighted imaging data, the following subjects' DWI data were removed from the pipeline (CC110033, CC110037, CC110056, CC110101, CC120008, CC120376, CC220901, CC320321, CC320698, CC321880, CC412004, CC420167, CC420435, CC420566, CC510256, CC510434, CC520083, CC620451, CC620567, CC710518, CC710551, CC721052, CC721377).

Another DWI Subject Removed

After quality control of the diffusion weighted imaging data, the following subjects' DWI data were removed from the pipeline (CC221585).

Correction of raw MEG BIDS

Owing to a file-naming error, all participants had their raw MEG BIDS files named "proc-sss" - this filename component now removed. Note this only happened to the new BIDS reformatting done on 11/04/2019 (not pseudo-BIDS format released previously).

Correction of multiple raw BIDS EPI runs

A subset of 33 participants had multiple runs in their raw BIDS EPI directories (for Rest, SMT and Movie tasks) - now removed and replaced with a single file (because only one run per task). Note this only happened to the new BIDS reformatting done on 11/04/2019 (not pseudo-BIDS format released previously).

Correction of multiple raw BIDS FieldMap runs

A subset of 33 participants had multiple runs in their raw BIDS FieldMap directories (for Rest, SMT and Movie tasks; like for EPI in issue above) - now removed and replaced with three files: sub-CC*_fmap.nii.gz is the phase map, sub-CC*_run-01_fmap.nii.gz is the magnitude map for short TE, and sub-CC*_run-02_fmap.nii.gz is the magnitude map for long TE. Note this only happened to the new BIDS reformatting done on 11/04/2019 (not pseudo-BIDS format released previously).

Removal of 5 subjects

Five subjects who only had a T1 image were removed from the dataset (CC210174, CC410447, CC610096, CC720330, CC710501)

Note lack of onset times for sensorimotor fMRI session in 4 subjects

Owing to a stimulus log problem, the onset .tsv files are missing for 4 subjects (CC120065, CC221755, CC410222, CC610146). Note however that event onsets are common across all subjects, bar a random ITI jitter of ~0.5s, so another subject's onsets could be used to analyse these 4 subjects if wished.

Echoes previously missing for movie fMRI task

Owing to a BIDS conversion problem, images for some echoes for 7 subjects were missing: CC223115, CC310400, CC320379, CC321053, CC410032, CC420720, CC510354. They have now been added.

Atypical fMRI acquisition parameters for movie task

5 subjects had different voxel resolutions due to acquisition errors. This is correctly noted in BIDS meta-files, but is mentioned in case important. Specifically, the common resolution is 3.00mm x 3.00mm x 4.44mm x 2.47s, but for CC310008, CC410040, CC510115, CC610028, CC710099 are 3.00mm x 3.00mm x 3.85mm x 2.47s.

DWI data recovered for 3 subjects, but removed for one

DWI data from 4 subjects were previously from aborted runs. Three of these have now been replaced with correct data (CC320321, CC420167, CC420566), but the data from the fourth (CC420435) have now been removed.

Epi-movie data for CC110037 were a duplicate of CC110033

The raw NII data in BIDS directory for subject CC110037 have now been corrected.

All MTI data updated with better de-facing (previous de-facing removed some of ATL)

The MTI data in BIDS directory for all subjects have now been corrected.

88Hz artifact in MEG data

There is a spike in power at 88Hz (and higher harmonics) in all the MEG datasets, which the manufacturer says is a complex aliasing of the HPI signals (even though they are energised >300Hz). It is a sharp peak, should not affect within-subject analyses, but could affect between-subject analyses if subjects have different HPI positions, possibly causing slight shifts in peak frequency.

A few erroneous events in MEG sensorimotor and passive tasks

All stimulus events corrected, and response events also added. Some subjects with incomplete or corrupt data removed. See README.txt in BIDSep folder for more information.

MEG BIDS format updated and MRI fiducials corrected.

BIDS files recreated using MNE-BIDS, during which it was noticed that fiducial info in the anat/sub-_T1w.json files were incorrect (identical for all subjects) - now corrected.

MEG derivatives (maxfiltered versions) updated.

Latest version (2.12) of Maxfilter re-run (within AA) on new MEG-BIDS structure above, resulting in slightly different data in derivatives folder and some path changes. Some subjects recovered. See details in README.txt in the BIDS root directories.

DWI preprocessing non-optimal.

Please be cautious using the "preproc_DWI" or "preproc_DTI" images, because their preprocessing is sub-optimal. For example, they have not been motion-corrected. Preprocessing of DWI data has improved over the intervening 8 years, and we hope to soon release improved versions. In the meantime, we recommend downloading the raw data and preprocessing yourself.

Bad DWI data.

Participant meta-data (participants.tsv) now also contains additional columns for quality control (DWI_stripe_index, DWI_stripe_index_flag) described in Neto-Henriques, R., Cam, C., & Correia, M. M. (2016). Reducing inter and intra-volume instabilities on diffusion-weighted data for ageing studies, Annual Meeting of the Organization for Human Brain Mapping. You should be wary of data from participants with a DWI stripe index > 0.1 (caused by intra-volume motion that not addressed by inter-volume motion correction like that described in previous issue above.

Poor contrast T1.

CC620359 has poor GM/WM contrast on their T1. Their other data look fine (including their later T1 in Stage 3), suggesting it was a problem with shimming for this one session. This causes problems for FreeSurfer, though SPM12 segmentations seem reasonable. Thanks to Christina Stier for detecting this.

Extraneous tab characters and participants.

The BIDS participants.tsv files for MEG rest, smt and passive tasks had an extraneous tab character at the end of each line, which caused problems for some functions that read these types of files. Also, the participants.tsv file for the passive MEG task had 9 extraneous (repeated) participants ate end. Now corrected.