Parametric mapping of scaled fitting error in dynamic susceptibility contrast enhanced MR perfusion imaging

The purpose of this study was to examine the benefits of routine generation of a parametric image of scaled curve fitting errors in the analysis of dynamic susceptibility contrast enhanced MR perfusion imaging. We describe the scaled fitting error (SFE), which reflects the magnitude of potential errors in the estimation of perfusion parameters from dynamic susceptibility contrast enhanced studies. The SFE is the root-mean-square error between the observed values in the time course of change of effective transverse relaxation rate (ΔR2*(t)) in tissue and the theoretical values derived by Γ variate curve fitting, scaled with a simple function related to the area under the fitted Γ variate curve. The SFE was tested using Monte Carlo simulation and by observations in normal volunteers and patients. This demonstrated that the SFE was linearly related to uncertainties in calculation of the values of relative cerebral blood volume (rCBV) and relative meant transit time (rMTT). High spatial resolution SFE maps were obtained in all volunteers and patients. In normal brain, SFE was consistently higher in white matter than in grey matter. In 54/85 patients with neurodegenerative or vascular brain disease, SFE maps showed focal areas with high values owing to poor signal to noise ratio in ΔR2*(t). Increased SFE was also found in 11/54 brain tumours owing to loss of conformance of ΔR2*(t) to the Γ variate function. SFE mapping is simple to implement and the computational overhead is negligible. It is concluded that parametric maps of SFE allow visual and quantitative comparison of fitting errors with the theoretical Γ variate model between anatomical regions and provide a quality control device to rapidly assess the reliability of the associated rCBV and rMTT estimations.