Encrypted login | home

Program Information

Uncertainty Analysis of Pharmacokinetic Modeling in Dynamic Contrast-Enhanced MRI Evaluated with MMID4 Simulation

no image available
C Liu

CY Liu1*, YP Liao2, YS Lin1,3, SC Chin4, HL Liu1,4, (1) Chang Gung University, Taoyuan, Taiwan, (2) Taipei Medical University- Shuang Ho Hospital, New Taipei City,Taiwan(3) Chang Gung Memorial Hospital-Keelung,Taiwan(4) Chang Gung Memorial Hospital-Linkou,Taoyuan, Taiwan

SU-E-I-18 Sunday 3:00PM - 6:00PM Room: Exhibit Hall

Purpose:
Uncertainty analysis including accuracy and precision evaluation is important for the pharmacokinetic modeling in dynamic contrast-enhanced (DCE) MRI. Few studies had attempted to assess the precision of the commonly used modified Tofts and Kermode (mTK) and the adiabatic approximation to the tissue homogeneity (AATH) models using neutrally simulated data. The study aimed to investigate the precision of these two models using the time curves simulated by multiple path, multiple tracer, indicator dilution, 4 region (MMID4) model, and perform comparisons in patients with nasopharyngeal carcinoma (NPC).

Methods:
Tissue uptake curves were simulated using the MMID4 model, by adjusting input parameters including blood flow, capillary permeability surface area product, interstitial volume (Ve) and capillary plasma volume(Vp), using literature values representing breast tumor and meningioma. Four levels of Gaussian noise, corresponding to SNRs of 100, 50, 20 and 10, were added to the signal time curves, with 1000 iterations each. A literature-based arterial input function was applied to the model. For NPC datasets, DCE-MRI was performed at a 3T clinical scanner using a 3D-SPGR sequence (TR/TE/FA=4.9ms/1.3ms/30degrees, 60 dynamics, sampling interval=3.9s).

Results:
Our simulation showed that the AATH provided better accuracy than the mTK model. From the results of the NPC patients, the AATH gave smaller transfer constant (Ktrans) and Ve, and larger Vp values, as compared to mTK model, which agreed with the simulation data. When comparing the coefficient of variance of common parameters by both models, the mTK resulted in superior precision than AATH, and greater CV was found with smaller SNR in all cases.

Conclusion:
Using the MMID4 model for data simulation, this work found that the AATH model resulted in more accurate but less precise estimates of physiological parameters as compared to the mTK model. Results from the analysis of clinical NPC data were consistent with the computer simulation.

Contact Email: