3-D Perfusion Imaging Using Principal Curvature Detection Rendering.

Three-dimensional contrast-enhanced ultrasound (CEUS) imaging presents a clear advantage over its 2-D counterpart in detecting and characterizing suspicious lesions as it properly surveys the inherent heterogeneity of tumors. However, 3-D CEUS is also slow compared to 2-D CEUS and tends to undersample the microbubble wash-in. This makes it difficult to resolve the feeding vessels, an important oncogenic marker, from the background perfusion cloud. Contrast-enhanced Doppler is helpful in isolating this conduit flow, but requires too many pulses in conventional line-by-line beamforming design. Recent breakthroughs in plane-wave imaging have greatly accelerated the volumetric imaging frame rate, but volumetric Doppler angiography still remains challenging when considering real-time limitations on the Doppler ensemble length. In this work, we demonstrate the feasibility of volumetric CEUS angiography subjected to real-time imaging constraints. Namely, we show how principal curvature detection can significantly improve 3-D rendering of relatively noisy ultrasound angiograms without degrading the spatial resolution while subjected to a reasonable Doppler ensemble size. Singular value decomposition is also shown to be capable of identifying the quasi-stationary capillary perfusion.