New sampling and filtering approaches to label-free ultrasonic perfusion imaging

Abstract

Color-flow and power Doppler imaging are safe, low-cost, non-invasive techniques for assessing blood flow throughout the body. Imaging microvascular flows and perfusion generally requires injectable contrast agents to separate tissue clutter from blood echoes. Our innovation is to modify echo acquisition to expand data dimensionality and increase sensitivity of non-enhanced PD imaging to very slow, disorganized flows. To separate perfusion from other echo signals, we developed a higher-order singular value decomposition filter that appropriately reduces data dimensionality. Effective elimination of clutter and noise results improved perfusion specificity. Our methods are applied to existing commercial US instruments, which makes them immediately valuable to medical practice. Our methods were previously tested at 24 MHz in a study involving muscle ischemia and highly-vascular melanoma lesions in mice. This study translates the methods to lower frequencies (5-10 MHz) for human trials. We developed a dialysis-cartridge device using TM-blood to quantifying very slow flow in the presence of tissue-like clutter and noise. We calibrated our technique to quantify the sensitivity, dynamic range, and spatial resolution of PD imaging. Results indicate an ability to separate flow speeds as low as 0.3 mm/s from clutter signals.