Direct estimation of tracer‐kinetic parameter maps from highly undersampled brain dynamic contrast enhanced MRI

Abstract

Purpose To develop and evaluate a T1-weighted dynamic contrast enhanced (DCE) MRI methodology where tracer-kinetic (TK) parameter maps are directly estimated from undersampled (k,t)-space data. Methods The proposed reconstruction involves solving a non-linear least squares optimization problem that includes explicit use of a full forward model to convert parameter maps to (k,t)-space, utilizing the Patlak TK model. The proposed scheme is compared against an indirect method that creates intermediate images via parallel imaging and compressed sensing prior to TK modeling. Thirteen fully-sampled brain tumor DCE-MRI scans with 5 sec temporal resolution are retrospectively undersampled at rates R=20, 40, 60, 80, and 100 for each dynamic frame. TK maps are quantitatively compared based on root mean-squared-error (rMSE), and Bland-Altman analysis. The approach is also applied to four prospectively R=30 undersampled whole-brain DCE-MRI data sets. Results In the retrospective study, the proposed method performed statistically better than indirect method at R≥80 for all thirteen cases. This approach provided restoration of TK parameter values with less errors in tumor regions-of-interest, an improvement compared to a state-of-the-art indirect method. Applied prospectively, the proposed method provided whole-brain high-resolution TK maps with good image quality. Conclusion Model-based direct estimation of TK maps from k,t-space DCE-MRI data is feasible and is compatible up to 100-fold undersampling.

Spatially regularized estimation of the tissue homogeneity model parameters in DCE‐MRI using proximal minimization

“…These are for example the wavelet transform as in Ref. favoring piecewise‐polynomial functions or the total generalized variation favoring piecewise linear functions. However, in our experience, the real‐life difference between them is not large.…”

“…Although using spatial priors is usual in image reconstructions including MRI, they have been used only occasionally in the DCE‐MRI analysis. To the authors’ knowledge, it has been used by only a few groups . The priors in the mentioned papers are based on image gradients of perfusion maps except for using a wavelet transform and using the difference of the image from its denoised variant.…”

“…All three mentioned transformations generate so‐called feature images on which a metric is computed to express, by one value, the spatial consistency of the voxels. The metrics used range from the smoothing norm to an approximation of the edge‐preserving norm . The minimization of these spatial‐regularization criterial functions is problematic because the voxels cannot be processed independently as in the case without the regularization.…”

“…Compared to where an approximate norm was used, in the present paper we derive a solution to the strict sparsity‐inducing norm regularizer operating on image gradients, that is, the total variation regularization. This reduces smoothing of edges between different tissues.…”

“…To the authors' knowledge, it has been used by only a few groups. [18][19][20][21][22][23][24] The priors in the mentioned papers are based on image gradients of perfusion maps except for 24 using a wavelet transform and 23 using the difference of the image from its denoised variant. The denoising is guided by the input DCE-MRI sequence.…”