Rapid Motion-Robust and Easy-to-Use Dynamic Contrast-Enhanced MRI for Liver Perfusion Quantification

Project Summary The broad objective of this application is to develop a rapid motion-robust and easy-to-use dynamic contrast- enhanced magnetic resonance imaging (DCE-MRI) framework for liver perfusion quantification and to evaluate its performance in quantitative assessment of hepatocellular carcinoma (HCC), the most prevalent primary malignancy in the liver. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) using gadolinium-based contrast agents is currently a cornerstone for identifying and characterizing hepatic lesions, including HCC. However, the current clinical use of liver DCE-MRI is limited to visual assessment of the pattern of perfusion in 3-4 multiphasic images (arterial, venous and delayed phases), and these images are routinely acquired during multiple breath holds. DCE-MRI also has the potential for quantitative assessment of perfusion kinetics, which can provide a deeper insight into the tumor microenvironment for non-invasive characterization of different histological features of the tumor, such as tumor angiogenesis and aggressiveness. This is particularly relevant for HCC, which is typically diagnosed based on imaging without pathological confirmation from invasive biopsy. Unfortunately, conventional liver perfusion MRI techniques suffer from a number of important limitations that restrict its clinical implementation, including (1) slow imaging speed, (2) limited spatiotemporal resolution, (3) sensitivity to motion artifacts, and (4) time-consuming quantitative perfusion analysis. Meanwhile, the need for pre-contrast T1 mapping to convert MR signal to gadolinium concentration further complicates the already-cumbersome imaging workflow. These challenges and underlying complexity have all led to non-reproducible performance of liver perfusion MRI and have significantly diminished its ultimate clinical utility. In this project, we propose to develop new rapid MRI techniques combining novel motion-robust sampling strategies and advanced reconstruction models to address these challenges. The new imaging techniques will enable motion-robust 3D T1 mapping with whole-liver coverage for efficient estimation of contrast concentration and free-breathing DCE-MRI of the liver with high spatiotemporal resolution. We will also incorporate state-of-the-art methods in deep learning to further improve imaging performance, to reduce reconstruction time, and to substantially simplify perfusion quantification. These new technical developments will be integrated into a new liver perfusion MRI framework, which will be translated into the clinical setting for assessment of HCC in an exploratory clinical study. The overall hypothesis is that with the new imaging framework developed in the project, robust high spatiotemporal resolution perfusion MRI of the liver can be achieved under free breathing, and absolute quantification of liver perfusion can be performed without user- interaction. Given the rapidly rising incidence and substantial burden of HCC in the United States, successful completion of this project would enable significant progress towards improved characterization and management of HCC and other liver diseases with high clinical impact.