Ultra-high resolution imaging and artery-vein separation of blood pool contrast-enhanced MRA

Contrast-enhanced magnetic resonance angiography (CE-MRA) is used for diagnosing atherosclerotic peripheral vascular occlusive disease (PVOD). Current methods of CE-MRA use conventional gadolinium (Gd) contrast agents to acquire images of the vasculature during the arterial phase. Accurate timing of the contrast arrival is required and acquisition times are limited due to the short contrast-enhancement window. A new type of Gd-based contrast agent, termed blood pool, has prolonged intravascular half-life and higher relaxivity than conventional Gd contrast agents. Image acquisition is no longer restricted to the dynamic phase. Longer scan times could be utilized for high resolution imaging, which would allow better visualization of the peripheral vasculature and diagnosis of vascular disease. This feature is useful in the lower extremity, where vessels are small (1-6mm in diameter). High resolution imaging occurs in the equilibrium phase (EP), where both arteries and veins are enhanced. Consequently, it is pertinent to distinguish and separate arteries and veins so that the arterial and venous vasculature can be visualized clearly.; Methods were developed to obtain ultra-high resolution images of the entire vasculature using the blood pool agent Vasovist and to separate those images into the arterial and venous vasculature. First, a model was developed to find optimal MR sequence parameters that maximize contrast-to-noise (CNR) for ultra-high resolution equilibrium phase imaging. An in-plane resolution of 0.5 mm was found to provide the best trade-off between the spatial resolution needed to distinguish arteries from veins and the CNR needed for artery-vein separation and detecting arterial disease.; An algorithm to segment distinct arterial and venous vasculatures from the EP MRA images was developed. The algorithm uses a novel feature image that incorporates vessel shape. This feature image can be applied in both Markov random field and level set methods. The algorithm is fast and requires only a small amount of user interaction to initialize and is hence practical for use in a clinical workflow. An image viewing strategy was also devised to alleviate the impact of imperfect segmentation.; The entire process from image acquisition through the clinical presentation of images was evaluated for effectiveness in diagnosing PVOD. The segmented EP images compared favorably to the dynamic phase images with high sensitivity and specificity, and significantly improved venous suppression, SNR, and vessel clarity.