Dear Hannah,
Glad you like it. 7-18y are old enough not to have any problems with T1-weighted image segmentation. And also old enough to have issues with T1 time of blood. So the problem might be unrelated to age.
One possible explanation is the PLD. 600 ms for example is too short. It would thus be good to examine the images visually to be sure that all artifacts are properly excluded and that all images show nice contrast in GM and not just macrovascular artifacts in vessels. You can go to root/derivatives/ExploreASL/Population/ASLCheck and browse through the DICOMs.
Another problem might be general scaling. For that, I would need to know a bit more about the data - sending an example ASL4D.json would be nice (since that doesn't contain any personal information). It then depends on scanner vendor, what might have went wrong.
best,
Jan