Premium
SU‐E‐I‐18: Variability of Physiological Parameters Estimated by AATH and MTK Models in DCE‐MRI
Author(s) -
Liu C,
Liao Y,
Liu H
Publication year - 2012
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4734733
Subject(s) - mathematics , mean squared error , root mean square , homogeneity (statistics) , dynamic contrast enhanced mri , square root , statistics , algorithm , physics , magnetic resonance imaging , geometry , medicine , radiology , quantum mechanics
Purpose: To assess the accuracy and precision of physiological parameters estimated by the modified Tofts and Kermode (mTK) model and the adiabatic approximation to the tissue homogeneity (AATH) model in dynamic contrast enhanced (DCE) MRI. Methods: Computer simulations were performed using the arterial input function (AIF) from Parker et al., and the multiple indicator, multiple path, indicator dilution 4 region (MMID4) model for simulating the tissue concentration time curve. A set of physiological parameters, simulating a breast tumor, was selected as the input to the MMID4. The output was then converted to signal time curve using SPGR signal equation. 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. These curves were then transferred back to concentration time curves on which mTK and AATH models were applied for fitting. For AATH model the mean transit time (tau) was held as a fixed parameter ranged from 2 to 10 s, and the final solution was selected based on minimizing the root mean square error. Results: For common parameters obtained by both models (Ktrans, Ve, Vp), AATH provided better accuracy, which agreed with the literatures, whereas mTK resulted in superior precision. As expected, greater coefficient of variation (CV) was found with smaller SNR in all cases. For the parameters that could only be obtained with the AATH model, the CVs were less than 10% for F, E and PS, and around 20% for tau, at the SNR level of 20. Conclusions: Using the MMID4 model for data simulation, this work found that AATH model was more accurate but less precise for estimating physiological parameters as compared with mTK model. The AATH model was able to separate blood flow and permeability estimates with reasonable errors and variations.