Advancing the Feasible Microbubble Concentration in Super-Resolution

Abstract

For the reconstruction of the microvasculature from contrast-enhanced ultrasound sequences with ultrasound localization microscopy, it is neccessary that a vessel is flown through by at least one microbubble (MB). Therefore, the reconstruction degree can be increased by either long acquisition times or by using high MB concentrations. However, in clinical applications the measurement times are usually limited. Thus, an adequate method for the detection of the MB is essential that ensures an accurate localization of single MB even in case of overlaying point-spread functions (PSFs) when using high MB concentrations. Therefore, we investigated the performance of sparsity based ultrasound hemodynamic super-resolution (SUSHI), and also expanded it to depth dependent PSFs. We applied the method to varying MB concentrations, analyzed different implementations of SUSHI and compared it to standard detection methods (Gauss detection and the centroid detection). The sparsity driven super-resolution with depth dependent PSFs showed the highest sensitivity for high MB concentrations. The Gauss detection yielded the lowest error rates. The centroid detection failed with increasing MB concentration.