A synthetic data set for validation of tracer kinetic modelling and model-driven registration in DCE-MRI

G. A. Buonaccorsi, G. J. Parker Imaging science and Biomedical Engineering, University of Manchester, Manchester, Manchester, United Kingdom Introduction In clinical quantitative dynamic contrast enhanced MRI (DCE-MRI), it is difficult to verify the accuracy of estimates of kinetic model parameters such as K, ve and vp, as we do not know the ‘true’ parameter values (ground truth). The problem is exacerbated when we apply motion-correction algorithms in addition to model fitting, as we need to validate both the image registration procedure and the fitting process. Therefore, as an initial step towards a complete validation framework, we have developed a synthetic DCE-MRI data set (software phantom) that incorporates the effects of contrast agent accumulation and washout, translational motion, and noise. Using this phantom and tracer kinetic modelling, we show that a recently introduced kinetic model-driven registration algorithm is able to recover accurate parameter values from motion-corrupted synthetic data.