Quantitative Evaluation of Temporal Sparse Regularizers for Compressed Sensing Breast DCE-MRI

We quantitatively evaluate temporal sparse regularizers for breast DCE-MRI data under standard compressed sensing schemes. We consider five temporal regularizers on 4.5x retrospectively undersampled Cartesian in vivo breast DCE-MRI data, namely Fourier transform (FT), Haar wavelet transform (WT), total variation (TV), second order total generalized variation (TGV$$$_{\alpha}^{2}$$$) and nuclear norm (NN). Both signal-to-error ratio and concordance correlation coefficients of the derived pharmacokinetic parameters $$$K^{\text{trans}}$$$ (volume transfer constant) and $$$v_\mathrm{e}$$$ (extravascular extracellular volume fraction) are estimated. Results show that NN produces the lowest image error while TV/TGV$$$_{\alpha}^{2}$$$ produce the most accurate pharmacokinetic parameters.