The Imagen database contains data collected and processed by the Imagen consortium from over 2000 adolescents and their parents. It includes demographics, neuropsychological assessments, medical questionnaires, MR neuroimaging and genomics. Data have been collected over a period of 10 years in 8 recruitment centers and over 4 successive time points: baseline at age 14 (BL), follow-up 1 at age 16 (FU1), follow-up 2 at age 19 (FU2) and follow-up 3 at age 22 (FU3).
To access the Imagen data, please refer to the data access policy.
We have written specific software and scripts for data collection, quality control, processing and publishing. We have published and maintain most of this software in GitHub.
We have released successive versions of the Imagen dataset:
- The seminal European project published data from baseline (BL) and the first follow-up (FU1) on a customized XNAT server (https://imagen.cea.fr). We don’t have detailed versioning information for these data releases and part of the software used to manage the dataset is not available any more, being partially proprietary or too complex to be maintained and published.
- For follow-ups 2 and 3 (FU2 and FU3), we moved as much as possible to open source software and scripts. Data have been released on a new server (https://imagen2.cea.fr) in well-defined versions and we have kept a log of these successive versions (numbered 2.*).
This is the first server that deliver the complete BL, FU1, FU2 datasets. Individual analysis are available regarding (i) task-based and resting state fMRI, (ii) FreeSurfer, NewSegment and SPM analysis for the anatomical data and (iii) pre-processing for the DW images. When available additional documentation was added.
- Improvment of the brainomics2 model efficiency (inline introduced),
- All pre-processed data for the task-based fMRI (NeuroSpin) and rs-fMRI (the latter processed by H Walter team – La Charité Hospital Berlin),
- All first line analysis data (processed NeuroSpin)
- All Freesurfer-segmentation data (processed and compiled NeuroSpin)
- Improvement of the registration tool.
We have detected the following errors in this dataset, to be fixed in the next release:
- The ADNI MPRAGE and short MPRAGE are both identified as ADNI MPRAGE.
- The following subjects had been swapped at BL and FU1, therefore disregard BL and FU1 data from these subjects:
- The following subjects had been swapped at FU2, therefore disregard FU2 data from these subjects:
- The FU2 imaging data of the following subjects is incorrect, therefore disregard FU2 imaging data from these subjects:
- Some Psytools questionnaires are missing and many Psytools questionnaires do not provide all the expected variables. As a workaround, download up-to-date and properly derived Psytools questionnaires as CSV files using SFTP.
- Dawba questionnaires use a mixture of dots and commas as decimal separators. As a workaround, download up-to-date Dawba questionnaires as CSV files using SFTP.
Only internal modifications performed, and very few new data added :
- major release of the data model (brainomics2)
- a few task-based fMRI pre-processed data for BL/FU2.
We made several important modifications :
- All user accounts imported from the old XNAT server were erased. All users are asked to create a new account to access the data.
- Upgrade questionnaire display and download.
- Correction in questionnaires: inconsistencies between float number format (e.g. “3,14” and “3.14”) have been corrected
CubicWeb framework: upgrade from 3.19.6 to 3.20.9
Fixed indexing issue for some questionnaires (ESPAD_CHILD_FU, ESPAD_CHILD_FU2, TLFB_FU2, VIDGAME_CHILD_FU2, CIDS_PARENT_FU2). Expression and methylation data are now available.
We added: the whole sets of questionnaires, imputed data, CNV data. The expression and methylation data will be released in a patch version. With this version, it is possible to reach some documentation attached to the data: the access to this documentation is context dependent. It must be noticed that this documentation is not complete and we will appreciate your feedback on it and contributions.
Version 2.3 exhibits some capabilities of the final renovated version. In particular:
- navigation according to natural pivotal concepts (subjects, exams),
- faceting functions that enables to refine intuitively the current
- statistics functions in the context of the view,
- access to the data download mechanisms,
- access rights are operating and enforced.
Some genetics were added.
All known subjects in the study (enrollment of subjects only occurs during BL) are indexed, but no DICOM images, questionnaires or scores are available from the new server (https://imagen2.cea.fr). Questionnaires and scores will be released in upcoming versions.
Initial version(s) of the Imagen dataset, with BL and FU1 data only, released during the course of the seminal European project using a modified XNAT server (https://imagen.cea.fr).