A fast, effective filtering method for improving clinical pulsed arterial spin labeling MRI

Area of Science:

• Medical Imaging
• Neuroimaging
• Radiology

Background:

• Pulsed arterial spin labeling (PASL) MRI is crucial for assessing cerebral blood flow (CBF).
• Image artifacts in PASL can compromise diagnostic accuracy.
• Automated postprocessing methods are needed to improve image quality.

Purpose of the Study:

• To evaluate an automated postprocessing filter for PASL MRI.
• To assess its effectiveness in a large clinical cohort.
• To determine if it reduces artifacts and improves interpretability of CBF maps.

Main Methods:

• A mean and standard deviation-based filter was developed to remove outliers in PASL perfusion-weighted images.
• Filtered and unfiltered quantitative cerebral blood flow (CBF) maps from 200 clinical cases were compared.
• Four blinded raters assessed the interpretability and artifact levels of the CBF maps.

Main Results:

• The automated filter successfully salvaged studies with motion artifacts, system instabilities, and acquisition disruptions.
• Filtered CBF maps demonstrated significantly fewer artifacts compared to unfiltered maps (P < 0.05).
• Rater assessments confirmed improved interpretability of filtered PASL MRI perfusion images.

Conclusions:

• Automated postprocessing filtering effectively reduces artifacts in PASL CBF maps caused by patient motion and system instability.
• This technique enhances the diagnostic quality of clinical perfusion imaging.
• Selective averaging via filtering improves image quality without significant loss of perfusion signal-to-noise ratio.