Arterial input function calculation in dynamic contrast-enhanced MRI: an in vivo validation study using co-registered contrast-enhanced ultrasound imaging

Abstract

Developing a method of separating intravascular contrast agent concentration to measure the arterial input function (AIF) in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) of tumours, and validating its performance in phantom and in vivo experiments. A tissue-mimicking phantom was constructed to model leaky tumour vasculature and DCE-MR images of this phantom were acquired. An in vivo study was performed using tumour-bearing rabbits. Co-registered DCE-MRI and contrast-enhanced ultrasound (CEUS) images were acquired. An independent component analysis (ICA)-based method was developed to separate the intravascular component from DCE-MRI. Results were validated by comparing the time-intensity curves with the actual phantom and in vivo curves. Phantom study: the AIF extracted using ICA correlated well with the true intravascular curve. In vivo study: the AIFs extracted from DCE-MRI using ICA were very close to the true AIF. Intravascular component images were very similar to the CEUS images. The contrast onset times and initial wash-in slope of the ICA-derived AIF showed good agreement with the CEUS curves. ICA has the potential to separate the intravascular component from DCE-MRI. This could eliminate the requirement for contrast medium uptake measurements in a major artery and potentially result in more accurate pharmacokinetic parameters. • Tumour response to therapy can be inferred from pharmacokinetic parameters. • Arterial input function (AIF) is required for pharmacokinetic modelling of tumours. • Independent component analysis has the potential to measure AIF inside the tumour. • AIF measurement is validated using contrast enhanced ultrasound and phantoms.