Structural Correlation-based Outlier Rejection (SCORE) algorithm for arterial spin labeling time series.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Biomedical Engineering

Background:

• 2D arterial spin labeling (ASL) is crucial for non-invasive cerebral blood flow (CBF) measurement.
• Artifacts from outlier control-label pairs can compromise ASL data quality and interpretation.
• Existing signal processing methods may not optimally address these artifacts, impacting reliability and sensitivity.

Purpose of the Study:

• To introduce and validate Structural Correlation-based Outlier REjection (SCORE), a novel algorithm for artifact removal in 2D ASL data.
• To enhance the reliability and sensitivity of CBF measurements derived from 2D ASL.
• To improve the detection of disease-related changes, such as those in Alzheimer's disease (AD).

Main Methods:

• SCORE algorithm developed for outlier control-label pair removal in 2D ASL.
• Validation using 2D pulsed ASL data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database.
• Assessment of within-subject coefficient of variation (wsCV) in control subjects to evaluate retest reliability.
• Evaluation of sensitivity in distinguishing AD patients from controls based on regional CBF differences.

Main Results:

• SCORE, particularly SCORE+ (with preprocessing), reduced wsCV by up to 21% in gray matter and 39% in smaller regions of interest (ROIs).
• An average 50% increase in effect size for patient-control differences was observed in AD-sensitive ROIs.
• These improvements in effect size were statistically significant (P < 0.05) for most ROIs and methods.

Conclusions:

• SCORE and SCORE+ generate CBF maps with improved retest reliability in control subjects.
• The algorithms enhance sensitivity to pathological CBF effects, aiding in the distinction between controls and patients.
• SCORE represents a significant advancement in processing 2D ASL data for neuroimaging research and clinical applications.