ABCD Release 2.0.1 fMRI-cleanup README
Issue: Due to incorrect post-processing, all task and resting-state fMRI data obtained on Phillips scanners should be excluded from all analyses. This includes tabulated and minimally processed data. This issue does not affect other modalities (sMRI, dMRI) or raw DICOM data (Fast Track). In DEAP and RDS, Philips fMRI data has been removed. They will be corrected and made available with Data Release 3.0. The field map direction for these data was mistakenly flipped, which led to increased distortion in processed fMRI images. 1512 participants are affected by this issue, which is 13% of baseline MRI visits.
Recommendations: We recommend that all users re-run fMRI related analyses without the Philips fMRI data. For published studies, we encourage users to confirm findings without the Philips fMRI data and provide a corrigendum with the updated results.
Instructions: This R script is to parse the ABCD Release 2.0.1 fMRI tabulated data to remove subjects that have incorrect post-processing distortion correction.
- Open the fMRI-cleanup.R in R, RStudio or preferred text editor.'
- Edit the working directory (line 2: setwd) to point to a location that contains the ABCD Release 2.0.1 tabulated fMRI text files downloaded from NDA. The full list of affected fMRI instruments is list below. Instruments from previous releases may have different names:
abcd_betnet02.txt abcd_midabwdp01.txt abcd_midabwdp202.txt abcd_midasemdp101.txt abcd_midasemdp202.txt abcd_midasemp102.txt abcd_midasemp202.txt abcd_midr1bwdp101.txt abcd_midr1bwdp202.txt abcd_midr1bwp102.txt abcd_midr1bwp202.txt abcd_midr2semp102.txt abcd_midr2semp202.txt abcd_midsemp102.txt abcd_midsemp202.txt abcd_tfabwdp101.txt abcd_tfabwdp201.txt abcd_tfnbr1semdp101.txt abcd_tfnbr1semdp201.txt abcd_tfnbr2bwdp101.txt abcd_tfnbr2bwdp201.txt abcd_tfnbr2dp101.txt abcd_tfnbr2dp201.txt abcd_tfncr1bwdp101.txt abcd_tfncr1bwdp201.txt abcd_tfsstabwdp101.txt abcd_tfsstabwdp201.txt abcd_tfsstasemdp101.txt abcd_tfsstasemdp201.txt abcd_tfsstr1bwdp101.txt abcd_tfsstr1bwdp201.txt abcd_tfsstr1semdp101.txt abcd_tfsstr1semdp201.txt abcd_tfsstr2bwdp101.txt abcd_tfsstr2bwdp201.txt abcd_tfsstr2semdp101.txt abcd_tfsstr2semdp201.txt abcd_tmidr1semdp101.txt abcd_tmidr1semdp202.txt abcd_tnbasemdp101.txt abcd_tnbasemdp201.txt abcd_tr2bwdp01.txt abcd_tr2bwdp202.txt abcd_tr2semdp101.txt abcd_tr2semdp202.txt midaparc03.txt midaparcp203.txt midr2bwp102.txt midr2bwp202.txt mrirscor02.txt mrisst02.txt mrisstr1bw01.txt mrisstr1sem01.txt mrisstr2bw01.txt mrisstr2bwsem01.txt mrisstsem01.txt nback_bwroi02.txt nbackallsem01.txt nbackr101.txt nbackr1sem01.txt nbackr201.txt nbackr2sem01.txt
- This folder also requires the MRI Info instrument (abcd_mri01.txt).
- Edit the output format (line 3: output_format) to export as either a csv file "CSV" or the original text format "TXT". Run the R script. As the files are large it can take a few minutes per table. Updated tables are created in the working directory with a "new_" prefix. Ensure there is enough disk space to create a copy of all the data to be processed.
Alternative datasets available: Additionally, a secondary release of processed ABCD fMRI data not affected by this issue will be shared shortly in NDA collection #3165 (https://nda.nih.gov/edit_collection.html?id=3165). These data have been processed through the ABCD-BIDS pipelines (https://github.com/ABCD-STUDY/abcd-hcp-pipeline). This collection will provide minimally processed data, but not tabulated data.