Function and structure registration problem in dHCP data

Hi everyone,
I have checked the functional and structural image registration in a subset of the dHCP data (about 100 subjects) and found 8 subjects that may have issues with resting-state functional image and T2w image misalignment. The following figure shows several subjects with this issue:

Has anyone noticed this problem?

I also notice that the QC report provided by dHCP has some alignment indexes, but these subjects are okay. I’m confused why the QC report didn’t capture the problem. Maybe I’m missing some information in the QC report?

Thank you!
Mengyu Tian