Spatiotemporal decomposition methods for nanobubble contrast-enhanced ultrasound

Nanobubble ultrasound contrast agents (NBs) offer enhanced access to deep tumor tissue by extrapolation from vasculature. When combined with targeting moieties, shell-stabilized nanobubbles can outperform other agents used in molecular ultrasound imaging due to their small size. The visualization and decoupling of extravasated and intravascular NBs is critical to furthering nanobubble molecular imaging with ultrasound. Spatiotemporal processing of echo data, usually via singular value decomposition (SVD), enables enhanced decoupling of signals of stationary from moving scatterers and therefore provides a novel tool for analysis of NB extravasation into the parenchyma. Instead of trivial solutions, we have developed non-negative SVD (nnSVD) methods applicable to most commercial scanners without requiring the raw data. Additionally, our formulation allows addition of proper regularizers, such as sparsity, to further expand the decomposition dimensions. We validate these methods in phantom models and present how the suggested decompositions can distinctly decouple flowing from extravasated NBs in in vivo tumour models with enhanced permeability. Our results show the successful tracking of nanobubble accumulation in tumour tissue and can serve as a relative metric for molecular targeting of prostate cancer. Further exploration of these methods could be used to improve long-term imaging and drug delivery capabilities of NBs.