Anomalous ADNI Datasets

[neurolabusc]

The Alzheimer’s Disease Neuroimaging Initiative (ADNI) provides carefully curated datasets. It has been brought to my attention that a few of these datasets have issues. These reflect issues with the datasets, not dcm2niix. Until the ADNI database is updated, human intervention will be required to censor or fix these images. Possible solutions are provided, but users should use their own judgement.

019_S_4293 Series 901 fcMRI

• Issue: Instance number 4184 erroneously claims to "(0008,9208) CS [REAL]" whereas it and all other images are magnitude
• Solution: Use "gdcmanon --dumb --remove 0008,9208 ..." to remove tag
• Issue: Instance number 4184 claims different echo time (30.0000991821289) vs others (30)
• Solution: Use "-m y" to force merging despite variable TE

006_S_6441 Series 901 fcMRI

• Issue: Image with instance number 549 is missing
• Issue: Images with instance numbers 9387..9456 are missing. Note that instance numbers are generated in the slice then time order (s1t1,s1t2...s1tn,s2t1...s2tn....sytn) rather than time then slice order (s1t1,s2t1...syt1, s1t2...).
• Solution: Do not use this participant

006_S_6277 Series 801 fcMRI

• Issue: Only 8740 of the 9456 images are provided
• Solution: Do not use this participant

006_S_6243 Series 901 fcMRI

• Issue: Only 8626 of the 9456 images are provided
• Solution: Do not use this participant

[i3roly]

thank you for this information chris.

i have passed it on to the appropriate individuals as well.

[i3roly]

i should note the dates for the scans above to help userse.

019_S_4293 - 2018-06-15 006_S_6441 - 2019-08-19 006_S_6277 - 2018-03-28 006_S_6243 - 2019-04-01

the suggestion for 019_S_4293 did work (and well, i think)

however duplicating the 541st instance to 549 did not. it still outputs a skewed fMRI. any tips chris?

[neurolabusc]

Why not send me your attempt at a repaired instance 541.

By the way, I always find DICOM file names crazy, you can use the function dcm2niix -r y /path/to/dicom To rename an individuals DICOMs. It will sort them by date, series number and name files by instance number and instance UID.

-c

Chris Rorden, PhD Endowed Chair of Neuroimaging Co-Director, McCausland Center for Brain Imaging Department of Psychology University of South Carolina Columbia, SC 29208, USA phone +1 803 404 2573 email rorden@sc.edumailto:rorden@sc.edu web www.mricro.com

On Aug 11, 2020, at 4:42 PM, gagan sidhu notifications@github.com<mailto:notifications@github.com> wrote:

i should note the dates for the scans above to help userse.

019_S_4293 - 2018-06-15 006_S_6441 - 2019-08-19 006_S_6277 - 2018-03-28 006_S_6243 - 2019-04-01

the suggestion for 019_S_4293 did work (and well, i think)

however duplicating the 541st instance to 549 did not. it still outputs a skewed fMRI. any tips chris?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://protect2.fireeye.com/v1/url?k=5c62d5cd-02f4eac0-5c629b0c-86f331ced5b2-b1ccc826b4c1ffb3&q=1&e=fb26627c-64db-4274-b567-f13c9d739d99&u=https%3A%2F%2Fgithub.com%2Frordenlab%2Fdcm2niix%2Fissues%2F416%23issuecomment-672268625, or unsubscribehttps://protect2.fireeye.com/v1/url?k=ef21de7a-b1b7e177-ef2190bb-86f331ced5b2-27c7570218c4807c&q=1&e=fb26627c-64db-4274-b567-f13c9d739d99&u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACEEL57CZS43N6MAIKRHXZLSAGULJANCNFSM4P3KEUDQ.

[i3roly]

hi chris,

sent the link via email (it's sensitive data).

what i did was i opened the 541th instance (ADNI_006_S_6441_MR_Axial_fcMRIEYESOPENbr_raw_20190820110042952_541_S862695_I1214057.dcm) then copied it to ADNI_006_S_6441_MR_Axial_fcMRIEYESOPENbr_raw_20190820110042952_549_S862695_I1214057.dcm then i opened matlab and changed the dicom instance number to 549 for the copied file.

still getting the "Chris Rorden's dcm2niiX version v1.0.20200808 Clang10.0.1 x86-64 (64-bit MacOS) Found 9386 DICOM file(s) Slice positions repeated, but number of slices (9386) not divisible by number of repeats (195): missing images? Hint: expected 48 locationsWarning: Interslice distance varies in this volume (incompatible with NIfTI format). Using RWVSlope:RWVIntercept = 0.733333:0 Philips Scaling Values RS:RI:SS = 0.733333:0:0.00919335 (see PMC3998685) Convert 9386 DICOM as ./AxialfcMRI(EYES_OPEN)a (64x64x9386x1) Unable to equalize slice distances: slice order not consistently ascending: dx=[0 0 0 0 0 0 ... "

issue (tried with -m y as well)

[neurolabusc]

Updated comments on 6441 above to details that it also has missing slices.