Conversion ADNI to BIDSS I found issue
59 views
Skip to first unread message
SeungKyu Kim
Jun 4, 2021, 5:22:43 AM6/4/21
to Clinica
Hi Clinica!
I found bug occurred from Image ID & Series ID.
I adopted to improve this issue.
Remove this line and then return the next line.
For example,
subject : 137_S_1041
visit : M06
Original Image ID : 56100
Original Series ID : 33082
t1_paths Image_ID : 56099
t1_paths Series_ID : 33082
I was wondering why only this image can not find the exact ID.
(just 1 difference between image id & series id)
Notably, Other Image find correct paths.
How Can I fix this problem?
SeungKyu Kim
Jun 7, 2021, 8:45:18 AM6/7/21
to Clinica
I found the solution of this issue.
you have to remove not only this line,
but also this line.
and then return function original_image.
The data of images is from MPRAGEMETA. I think the function check_qc is something wrong.
I will also report this issue to GitHub.
2021년 6월 4일 금요일 오후 6시 22분 43초 UTC+9에 SeungKyu Kim님이 작성:
Omar Rifai
Jun 7, 2021, 10:12:30 AM6/7/21
to Clinica
Hi SeungKyu Kim,
Thank you for your interest in Clinica. Occasionally, the image ID saved in the t1_paths.csv file may not correspond to that of the image ID converted. This occurs when the converter finds a "Scaled" image (which is the most advanced processing step in ADNI) in the raw data. For convenience, we keep this "Scaled" image ID in t1_paths.csv but we actually convert the "N3" processed image for homogeneity with the more recent ADNI studies whereby scaling was abandoned.
The reason we keep the "Scaled" image ID is because it is the ID that can help us look for this image on the ADNI Website.
You can find a discussion around this issue in here: https://github.com/aramis-lab/clinica/issues/204
Meanwhile, this is the best compromise we have found currently.
Reply all
Reply to author
Forward
0 new messages