Achievable Localization Precision of Clinical 3D Ultrasound Localization Microscopy (ULM)

Abstract

For the translation of Ultrasound Localization Microscopy into clinical practice, it is essential to determine the localization precision of the microbubbles achievable with clinical 3D ultrasound devices before their first in vivo application. While a constant bias (accuracy) still allows for a shifted but correct reconstruction of the microvasculature, the achievable resolution is limited by the precision. Here, we propose an easy measurement setup and method for precision measurements that overcomes the known limitations of phantom measurements. With this, we compare the precision of two ultrasound matrix transducers (X5-1 and XL14-3) and different imaging sequences of the Epiq 7 US device. We also test different localization methods (Center of Mass, Gaussian Fit, and Radial Symmetry Centers). We show that the localization precision is mainly influenced by the imaging sequence and only slightly by the localization method. Interestingly, the XL14-3 yields worse precisions (7.8 μm to 51 μm) than the X5-1 (2.8 μm to 15.2 μm), despite its higher center frequency. Generally, the Center of Mass seems to be the most robust method for the localization of the microbubbles.