Sparsity transform k-t principal component analysis for accelerating cine three-dimensional flow measurements.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Fluid Dynamics

Background:

• Time-resolved 3D flow measurements are crucial for cardiovascular analysis but are limited by long acquisition times.
• Existing acceleration techniques, such as k-t methods, show promise for reducing scan times by leveraging spatiotemporal correlations.

Purpose of the Study:

• To develop an enhanced k-t principal component analysis (PCA) method for accelerating time-resolved 3D flow MRI.
• To improve reconstruction accuracy and enable higher spatial resolution in cardiovascular flow imaging.

Main Methods:

• Proposed an extension of k-t PCA utilizing signal differences between velocity encodings in three-directional flow measurements.
• Incorporated sparsity transform to further compact signal representation and improve reconstruction.
• Validated the method using simulated and in vivo data of the carotid bifurcation with 8-fold undersampling.

Main Results:

• Sparsity transform reduced velocity root-mean-square errors by 52% (CCA), 59% (ECA), and 16% (ICA) in simulated data.
• In vivo, errors in the common carotid artery (CCA) were reduced by 15% with sparsity transform.
• Achieved 0.8 mm isotropic spatial resolution with 8-fold undersampling in 6 minutes, revealing detailed helical flow patterns.

Conclusions:

• The proposed k-t PCA extension with sparsity transform effectively reduces undersampling artifacts and improves accuracy in 4D flow MRI.
• This method enables significantly faster acquisition at higher spatial resolutions, enhancing the visualization of complex blood flow dynamics.