ICA cleanup for improved SNR in arterial spin labeling perfusion MRI

Arterial spin labeling (ASL) is a non-invasive MRI modality that can provide insight in brain hemodynamics. One main limiting factor of ASL is its relatively low signal-to-noise ratio (SNR). New technical developments like 3D readouts and background suppression have improved SNR [1] and additional post processing steps including noise regression methods can further improve temporal SNR (tSNR) [2]. We hypothesize that Independent Component Analysis (ICA) should provide separation of physiological noise from signal and thus improving SNR and cerebral blood flow (CBF) quantification as has been shown for BOLD fMRI. Therefore, in this study, we evaluated the use of ICA to separate perfusion signal from noise in ASL data.