Non-linear fitting with joint spatial regularization in arterial spin labeling.

Area of Science:

• Medical Imaging
• Neuroimaging
• Quantitative MRI

Background:

• Multi-delay single-shot arterial spin labeling (ASL) provides crucial cerebral blood flow (CBF) and arterial transit time (ATT) maps.
• Low signal-to-noise ratio (SNR) in ASL imaging challenges standard quantification methods, leading to noisy estimates.
• Existing advanced techniques improve SNR but often introduce spatial blurring by incorporating prior knowledge.

Purpose of the Study:

• To develop and evaluate a novel joint spatial total generalized variation regularization method for ASL quantitative map estimation.
• To improve noise suppression and maintain sharpness in CBF and ATT maps derived from ASL imaging.
• To enhance the accuracy and reliability of quantitative ASL parameters.

Main Methods:

• Proposed a new estimation method employing joint spatial total generalized variation regularization on both CBF and ATT maps.
• Leveraged shared spatial features between CBF and ATT maps to enhance sharpness and reduce noise.
• Evaluated the method on synthetic phantom data, in vivo healthy volunteer data, and patient data with ischemic stroke.

Main Results:

• The proposed joint regularization method demonstrated superior performance compared to non-linear least squares fitting and a Bayesian inference-based algorithm.
• Achieved substantial increases in SNR for both CBF and ATT maps.
• Successfully maintained map sharpness and quantitative accuracy, outperforming reference methods.

Conclusions:

• The joint spatial total generalized variation regularization is an effective approach for improving ASL quantitative map estimation.
• This method offers enhanced SNR and sharpness without compromising quantitative accuracy, addressing key limitations of current ASL techniques.
• The proposed method shows significant potential for clinical applications, particularly in neurovascular imaging and stroke assessment.