Click here to


Are you sure ?

Yes, do it No, cancel

Calibration of Synthetic MRI Acquisition Parameters Through Information Theory Modeling

D Mitchell*, K Hwang , R Stafford , J Bankson , D Fuentes , UT MD Anderson Cancer Center, Houston, TX


(Sunday, 7/14/2019) 2:00 PM - 3:00 PM

Room: 221CD

Purpose: We model an adaptation of 3D QALAS (3D-quantification using an interleaved Look-Locker acquisition sequence with T2 preparation pulse) for 3D multi-parameter quantification in the brain, a novel technique based on a multi-acquisition 3D gradient echo sequence. Many existing parameter quantification methods used to generate ‘synthetic’ MRI contrasts require clinically unacceptable scan times, while fast methods typically have a narrow range of accuracy or require high SNR to obtain adequate estimates. We seek to employ information theory to address these two drawbacks by quantifying the information content of potential acquisitions and subsequently optimizing mutual information to allow selection of parameters which maximize synthetic MRI reproducibility.

Methods: A mathematical model of the 3D QALAS acquisition sequence represents the uncertainty in parametric map reconstruction and machine noise during typical acquisitions. A recursive conditional mutual information formulation quantifies information content of new measurements given previous acquisitions with independent parameters. A representative multi-contrast target phantom is used to estimate mutual information and predict optimal acquisition parameters. To test acquisition parameter optimization, two scans were performed on a System Standard Model 130 phantom (QalibreMD, Boulder, CO) with a 3T scanner (MR750, GE Healthcare, Waukesha, WI).

Results: Reconstruction uncertainty is measured by the standard deviation of parametric map values within the phantom elements. For M0, T1, and T2 maps, the standard deviation of reconstructed values is negatively correlated with mutual information. A reconstruction from an acquisition conditioned on a low-resolution pre-scan image also demonstrates reduced reconstruction variance compared to an image from an unconditioned acquisition.

Conclusion: This information theoretic analysis provides a metric by which to predict image noise and reproducibility prior to acquisition. Optimization of this metric has potential applications in corrective updates to acquisitions in real time, which could include updating pulse sequence parameters mid-scan to maximize information acquired within clinical constraints.

Funding Support, Disclosures, and Conflict of Interest: Research support was provided in part by GE Healthcare.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email