Influence of Signal to Noise Ratio on the Pharmacokinetic Analysis in DCE-MRI Studies

Recruiting the pharmacokinetic parameters estimated from noninvasive methods such as Dynamic Contrast Enhanced MRI (DCE-MRI) to evaluate or plan treatment procedure is widely interested in clinical practices. Interpretation of the DCE-MRI data are highly dependent to precision and accuracy of the estimated parameters. One of the most effective factors on the DCE-MR images and consequently on the contrast concentration profile is signal to noise ratio. This work focuses on the analytically evaluation of the noise effect on accuracy of the estimated PK parameters in DCE-MRI studies.

Tofts model as a popular pharmacokinetic model and model selection technique was used to simulate 3470 time curves of contrast concentration. Maximum likelihood estimator as a minimum variance unbiased estimator was recruited to estimate the PK parameters.  Eleven levels of signal to noise ratios (SNR= 5, 8, 10, 13, 15, 20, 25, 30, 35, 50, Noiseless) were added to the simulated CA concentration profiles. The PK parameters were estimated for 11 series data and then mean percentage error was calculated for estimated parameters.

The results indicate that the most sensitive parameters to the SNR of the DCE-MR images is inverse transfer constant. A SNR greater than 25 was found to be ensure a reasonable error (<5%) in all models parameters.

Clinical decision based on the DCE-MRI data analysis and estimated PK parameters needs a good image quality (SNR>25), an accurate and robust estimator and correct pharmacokinetic model selection.