Comparison of the kinetic parameters estimated with LLSQ and NLSQ methods in DCE-MRI

Abstract

Dynamic contrast enhanced (DCE) MRI plays an important role for quantitative assessment of tumor characterization using the kinetic parameters, such as volume transfer constant (Ktrans) and rate constant (kep). Nonlinear least square (NLSQ) method is often used to fit concentration time curves to kinetics models for obtaining kinetic parameters. However, one disadvantage of this method is time-consuming. The previous study demonstrated that liner least-squares (LLSQ) method may be a more accurate and fast method than NLSQ method in solving modified Tofts and Kermode model. In this study, we performed computer simulations to assess the accuracy of estimated the kinetic parameters for LLSQ and NLSQ methods in solving Tofts and Kermode model. In addition, the two methods were used to analyze the data from eight patients with brain tumors. At lower signal-to-noise (SNR), the accuracy of parameters estimated with LLSQ method was better than NLSQ method. Compare with NLSQ method, effect of varying Ktrans and kep on accuracy and precision of the kinetic parameters with LLSQ method was smaller. Besides, the calculation velocity of LLSQ method was seventeen times faster than NLSQ method. In clinical application, there were significant differences between LLSQ and NLSQ estimated kinetic parameters on the all patients with glioblastoma multiforme (GBM).