open-discussion > Issues with poor quality scans
Showing 1-2 of 2 posts
Apr 19, 2022 07:04 PM | Alicia Hilderley
Issues with poor quality scans
I have noticed that DTIprep is less sensitive for poor quality
scans (e.g., >30% bad volumes). Volumes that have visible
artifacts are not flagged as bad during slice-wise checking. I am
using default parameters. Should I adjust any parameters?
Thank you,
Alicia
Thank you,
Alicia
Apr 20, 2022 05:04 PM | Martin Styner
RE: Issues with poor quality scans
Originally posted by Alicia Hilderley:
The thresholds are based on the variance in the data (e.g. the variance on inter-slice correlation) and thus adjusts to the image quality settings. But, you can adjust the thresholds in the protocol. Increasing thresholds will lead to more generous QC and reducing them will lead to more strict QC.
I would start with the slice-checking threshold and reduce that one such that you get the desired results for your data.
If your artifacts are more motion related and you see a slice-by-slice offset when viewing the data in a slice orthogonal to the acquisition plane, then you could also adjust the interlace thresholds
Best
Martin
I have noticed that DTIprep is less sensitive
for poor quality scans (e.g., >30% bad volumes). Volumes that
have visible artifacts are not flagged as bad during slice-wise
checking. I am using default parameters. Should I adjust any
parameters?
Thank you,
Alicia
Thank you,
Alicia
The thresholds are based on the variance in the data (e.g. the variance on inter-slice correlation) and thus adjusts to the image quality settings. But, you can adjust the thresholds in the protocol. Increasing thresholds will lead to more generous QC and reducing them will lead to more strict QC.
I would start with the slice-checking threshold and reduce that one such that you get the desired results for your data.
If your artifacts are more motion related and you see a slice-by-slice offset when viewing the data in a slice orthogonal to the acquisition plane, then you could also adjust the interlace thresholds
Best
Martin