Removing the effect of SVD algorithmic artifacts present in quantitative MR perfusion studies

Area of Science:

• Medical imaging
• Neuroimaging
• Quantitative MRI

Background:

• Dynamic susceptibility contrast (DSC) MR perfusion imaging is crucial for quantitative cerebral blood flow (CBF) assessment.
• Standard singular value decomposition (sSVD) is a common deconvolution algorithm used in DSC MRI.
• Existing literature suggests sSVD-derived CBF estimates are sensitive to arterial-tissue delay (ATD).

Purpose of the Study:

• To investigate the diagnostic reliability of quantitative CBF measurements in DSC MRI.
• To evaluate the impact of arterial-tissue delay (ATD) on CBF estimates derived from sSVD deconvolution.
• To introduce and validate a reformulated singular value decomposition (rSVD) algorithm for improved CBF quantification.

Main Methods:

• Comparison of standard singular value decomposition (sSVD) and Fourier transform (FT) deconvolution algorithms.
• Simulation studies to assess the influence of ATD on CBF estimates.
• Validation using patient case studies with DSC MR perfusion data.
• Development and testing of a reformulated SVD (rSVD) deconvolution algorithm.

Main Results:

• Standard sSVD deconvolution demonstrates significant dependence of CBF estimates on ATD.
• Fourier transform (FT) deconvolution yields CBF estimates that are independent of ATD.
• The ATD-dependency of sSVD is identified as an artifact of its current implementation.
• The reformulated SVD (rSVD) algorithm shows improved robustness against ATD variations.

Conclusions:

• The diagnostic reliability of quantitative CBF measurements in DSC MRI is compromised by ATD-dependent sSVD algorithms.
• A reformulated SVD (rSVD) deconvolution algorithm provides more accurate and reliable CBF quantification.
• rSVD minimizes artifacts related to ATD, enhancing the diagnostic utility of DSC MR perfusion imaging.