Toward local arterial input functions in dynamic contrast-enhanced MRI

Area of Science:

• Medical Imaging
• Radiology
• Pharmacokinetics

Background:

• Accurate estimation of the arterial input function (AIF) is crucial for quantitative analysis in dynamic contrast-enhanced MRI (DCE-MRI).
• Traditional methods often rely on global or population-averaged AIFs, which may not accurately represent individual patient physiology.

Purpose of the Study:

• To present and validate a novel method for estimating the local arterial input function (AIF) using the alternating minimization with model (AMM) algorithm in DCE-MRI.
• To assess the impact of using local AIFs versus global or measured AIFs on pharmacokinetic parameter estimation.

Main Methods:

• A new method clusters DCE-MRI data into representative curves, feeding into the AMM algorithm to derive a parameterized local AIF and pharmacokinetic parameters.
• Computer simulations were employed to evaluate the accuracy of the AMM in estimating the true AIF based on input tissue curves and signal-to-noise ratio (SNR).

Main Results:

• Simulations demonstrated a power-law relationship between uncertainty in kinetic parameters, SNR, and input heterogeneity.
• Kinetic parameters derived using a local AIF were significantly different from those calculated with global AIFs (P < 0.005) or measured AIFs (P = 0.0).
• Local AIFs resulted in substantial mean lesion-averaged changes in K(trans) (+24%) and k(ep) (+13%), while global AIFs yielded smaller changes (+9% and +13%, respectively).

Conclusions:

• The use of local arterial input functions leads to significantly different kinetic parameter values compared to global estimations.
• The statistically significant improvement in model fit quality when using local AIFs suggests these estimates better reflect underlying tissue physiology.