Improving 3D-CINE tTV-regularized whole-heart MRI reconstruction

PurposeTo improve the image quality of 3D radial free-running MRI data of the heart through a deliberate and stepwise extension of the XD-GRASP reconstruction. MethodsFerumoxytol-enhanced cardiac free-running 3D-radial data were reconstructed using an XD-GRASP reconstruction improved by 4 new developments: motion-compensated temporal-Total-Variation (MC-tTV) regularization for 3D images, a new coil-sensitivity, a new k-space density compensation and a revisited conjugate-gradient-descent (with exact line search) for solving the least-square sub-problem of ADMM. The resulting images were compared quantitatively and qualitatively to reconstructions lacking some of the newly implemented measures. Also, the measurement of ejection-fraction by a threshold-based method on the new reconstruction was compared to a reference standard. ResultsThe new reconstruction significantly increased the sharpness of the right coronary artery (4% to 6%, p < 0.05) and the left anterior descending coronary artery (4% to 5% p < 0.05). It also increased blood-myocardium interface sharpness (between 20% and 25%, p < 0.05) and decreases spatial-Total-Variation in the blood-pool (13%, p < 0.05). The qualitative evaluation suggests better anatomical depiction of small structures using the new reconstruction. As compared to a reference standard method, ejection fraction could also be correctly evaluated. ConclusionCompressed sensing image reconstruction for 3D-radial free-running cardiac acquisition was successfully improved by including MC-tTV regularization, a new density compensation, a new coil-sensitivity and a revisited conjugate-gradient-descent with exact line search. Quantitative and qualitative quality metrics demonstrated significant improvement in image quality when using the new reconstruction, while extracted dynamic information compared favorably with the gold standard.