(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose:
Increase the temporal resolution of dynamic contrast enhanced MRI (DCE-MRI), by using small fractions of the total k-space dataset to reconstruct high temporal resolution images of the early phase of contrast media uptake. The method was verified by both simulations and mouse imaging.
Methods:
Enhancement-constrained acceleration (ECA) selects the portion of k-space acquired during the desired time intervals and initially zero-fills the rest of k-space. We reconstruct high temporal resolution images by solving a smoothness-constrained optimization problem. A positive smoothness penalty matrix penalizes the per-voxel discretized second time derivative in the image-domain. Minimization is solved by conjugate gradient descent. To ensure fidelity, only the k-space points that were not measured in each interval are adjusted to minimize the penalty.
A digital phantom of breast DCE-MRI data was used to verify the accuracy of the method. Then the method was tested in mice. Images of the femoral artery/vein were obtained using a 9.4 T MRI system (Bruker, Biospin) with a spoiled gradient echo sequence (nominal temporal resolution: 5 and 10 sec). Omniscan was injected I.V. K-space data were partitioned to achieve reconstructed temporal resolution of 1 second.
Results:
High temporal resolution images of the digital phantom reconstructed with ECA accurately reproduced the ground truth low temporal resolution images (Figure 1 and Figure 2) and provided additional information on the early phase of contrast agent uptake (Figure 3). To obtain the arterial input function (AIF) in mice, we measured the signal phase in the femoral artery. In Figure 4, ECA shows an AIF peak that was missed by standard images. This demonstrates that ECA recovers temporal features that are missed by standard Fourier methods.
Conclusion:
ECA significantly increases temporal resolution during the initial phase of contrast media uptake, when enhancement is very sparse. These early-enhancing features provide critical diagnostic information.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
