It is often difficult to accurately localize small arteries in images of peripheral organs, and even more so with vascular abnormality vasculatures, including collateral arteries, in peripheral artery disease (PAD). This poses a challenge for manually sampling arterial input function (AIF) in quantifying dynamic contrast-enhanced (DCE) MRI data of peripheral organs. In this study, we designed a multi-step screening approach that utilizes both the temporal and spatial information of the dynamic images, and is presumably suitable for localizing small and unpredictable peripheral arteries. In 41 DCE MRI datasets acquired from human calf muscles, the proposed method took <5 s on average for sampling AIF for each case, much more efficient than the manual sampling method; AIFs by the two methods were comparable, with Pearson's correlation coefficient of 0.983 ± 0.004 (p-value < 0.01) and relative difference of 2.4% ± 2.6%. In conclusion, the proposed temporospatial-feature based method enables efficient and accurate sampling of AIF from peripheral arteries, and would improve measurement precision and inter-observer consistency for quantitative DCE MRI of peripheral tissues.
Keywords: Arterial input function; Calf muscles; Connected component analysis; Magnetic resonance imaging; Peripheral artery disease.
Copyright © 2018 Elsevier Inc. All rights reserved.