Extracting CBF and ATT from One or Two Post-labeling Delays-based Arterial Spin Labeled Perfusion MRI via Deep Learning

Cerebral blood flow (CBF) and arterial transit time (ATT) can be quantified through fitting the arterial spin labeling (ASL) perfusion MRI signal acquired at different post-labeling delays (PLDs) into a kinetic model. Acquiring multiple-PLD ASL MRI needs exponentially prolonged total scan time compared to the single-PLD acquisition, making it highly sensitive to motions and impractical for clinical use. We proposed a deep neural network that can reliably estimate ATT and CBF maps from significantly fewer PLD ASL MRI acquisitions without image quality loss.