Thomas S.C. Ng^{1}, Ravi T. Seethamraju^{2}, and Ritu R. Gill^{1}

Adoption of quantitative clinical DCE-MRI remains limited given the challenges in accurate and precise estimation of kinetic parameters. Simulations are being increasingly used to guide protocol optimization1, but the relative effects of altering protocol variables are seldom considered. We used a modified K-CNR metric to quantify tumor Ktrans estimation. K-CNR provided a simple way to compare how input variables affect Ktrans output. The extended Toft’s model was shown to be robust for tumor relevant Ktrans. Lengthening baseline time can improve Ktrans estimation. Care must be taken when using nested model analysis; wrong model convergence can occur with non-optimized acquisition variables.

1. Artzi, M.; Liberman, G.; Nadav, G.; Blumenthal, D. T.; Bokstein, F.; Aizenstein, O.; Ben Bashat, D., Optimization of DCE-MRI protocol for the assessment of patients with brain tumors. Magn. Reson. Imaging 34 (9), 1242-1247.

2. Parker, G. J. M.; Roberts, C.; Macdonald, A.; Buonaccorsi, G. A.; Cheung, S.; Buckley, D. L.; Jackson, A.; Watson, Y.; Davies, K.; Jayson, G. C., Experimentally-derived functional form for a population-averaged high-temporal-resolution arterial input function for dynamic contrast-enhanced MRI. Magn. Reson. Med. 2006, 56 (5), 993-1000.

3. Barnes, S. R.; Ng, T. S. C.; Montagne, A.; Law, M.; Zlokovic, B. V.; Jacobs, R. E., Optimal acquisition and modeling parameters for accurate assessment of low Ktrans blood-brain barrier permeability using dynamic contrast-enhanced MRI. 2015, In prep.

4. Duan, C.; Kallehauge, J. F.; Bretthorst, G. L.; Tanderup, K.; Ackerman, J. J.; Garbow, J. R., Are complex DCE-MRI models supported by clinical data? Magn. Reson. Med. 2016.

5. Barnes, S. R.; Ng, T. S.; Santa-Maria, N.; Montagne, A.; Jacobs, R. E., ROCKETSHIP: a flexible and modular software tool for the planning, processing and analysis of dynamic MRI studies. Manuscript submitted for publication 2015.

6. Di Giovanni, P.; Azlan, C. A.; Ahearn, T. S.; Semple, S. I.; Gilbert, F. J.; Redpath, T. W., The accuracy of pharmacokinetic parameter measurement in DCE-MRI of the breast at 3 T. Phys. Med. Biol. 2010, 55 (1), 121.

7. Aerts, H. J.; Jaspers, K.; Backes, W. H., The precision of pharmacokinetic parameters in dynamic contrast-enhanced magnetic resonance imaging: the effect of sampling frequency and duration. Phys. Med. Biol. 2011, 56 (17), 5665-78.

8. Cramer, S. P.; Larsson, H. B. W., Accurate determination of blood-brain barrier permeability using dynamic contrast-enhanced T1-weighted MRI: a simulation and in vivo study on healthy subjects and multiple sclerosis patients. J. Cereb. Blood Flow Metab. 2014, 34 (10), 1655-1665.

9. Khalifa, F.; Soliman, A.; El-Baz, A.; El-Ghar, M. A.; El-Diasty, T.; Gimel’farb, G.; Ouseph, R.; Dwyer, A. C., Models and methods for analyzing DCE-MRI: A review. Med. Phys. 2014, 41 (12), 124301.

Fitted vs. simulated
Ktrans for (a) different SNR and (b) different time resolution. Decrease SNR
leads to less precise estimation while longer time resolution leads to less
accurate and precise estimation.

K-CNR analysis for different input parameters. K-CNR
analysis for varying (a) SNR, (b) time resolution, (c) acquisition time and (d)
baseline time demonstrates the differential effects upon Ktrans estimation by
each variable. All fits were performed with the ExTofts model. Increasing
baseline times improves Ktrans estimation more than acquisition time. Such
analysis would be helpful for protocol development.

Model convergence
using a nested model is dependent on input variables. Commensurate with K-CNR
analysis, appropriate model convergence in complex DCE-MRI models require
variables that maximize K-CNR (e.g. high SNR, short time resolution).

Wrong model
convergence choice results in inaccurate Ktrans estimation. Toft’s estimation
of the simulated data results in overestimation of Ktrans (a), which is
reflected in the K-CNR (b). Understanding how our input variables affect Ktrans
estimation using simulation analysis and K-CNR can guide appropriate analysis
and processing of acquired data.

K-CNR analysis allows
effect comparisons amongst multiple input parameters. Here, lower SNR (=15)
with short time resolution and longer baseline time was demonstrated to achieve
similar Ktrans estimation ability as a higher SNR but longer time resolution.
Taken together with results in Figure 4, this may influence our decision to
adopt a lower SNR protocol given its relative robustness to nested model
evaluation.