Development of Super-Resolution Sharpness-Based Axial Localization for Ultrasound Imaging
Development of Super-Resolution Sharpness-Based Axial Localization for Ultrasound Imaging
复制标题
DOI：
10.1109/access.2018.2889425
复制
发表时间：
2019
期刊：
影响因子：
3.9
通讯作者：
K. Diamantis;T. Anderson;J. Jensen;P. Dalgarno;V. Sboros
中科院分区：
文献类型：
--
作者：
K. Diamantis;T. Anderson;J. Jensen;P. Dalgarno;V. Sboros
超分辨率超声主要使用基于图像的方法来定位单个散射体。这些方法主要基于质心（COM）计算。基于锐度的定位是一种替代COM在轴向方向上的散射体定位。模拟超声点散射体数据（中心频率$f {0} = 7$ MHz，波长$\lambda = 220\，\，\mu\text{m}$）表明，归一化锐度法可以提供散射体轴向定位，精度低至$2\mu \text{m}$（$），这与通过常规成像可实现的相比是两个数量级的改进（$\approx \lambda $），并且与COM估计值（$\approx 10~\mu \text{m}$或$0.05\lambda $）相比提高了五倍。使用合成孔径实时超声系统采集的线靶数据进行实验，获得了类似的结果。所提出的方法的性能也被发现是一致的，在不同类型的超声波传输。在噪声存在的情况下，定位精度下降，但即使在很低的信噪比（SNR = 0 dB）下，定位精度的不确定性也不超过6 μ m，优于COM估计。该方法可以在图像数据中以及通过使用原始信号来实现。有人建议，信号衍生的定位应取代基于图像的等效，因为它提供了至少10倍的提高精度。
Super-resolution ultrasound mostly uses image-based methods for the localization of single scatterers. These methods are largely based on the center of mass (COM) calculation. Sharpness-based localization is an alternative to COM for scatterer localization in the axial direction. Simulated ultrasound point scatterer data (center frequency $f_{0} = 7$ MHz and wavelength $\lambda = 220\,\,\mu \text{m}$ ) showed that the normalized sharpness method can provide scatterer axial localization with an accuracy down to $2~\mu \text{m}$ ( $), which is a two-order of magnitude improvement compared to that achievable by the conventional imaging ( $\approx \lambda $ ), and a five-fold improvement compared to the COM estimate ( $\approx 10~\mu \text{m}$ or $0.05\lambda $ ). Similar results were obtained experimentally using wire-target data acquired by the Synthetic Aperture Real-time Ultrasound System. The performance of the proposed method was also found to be consistent across different types of ultrasound transmission. The localization precision deteriorates in the presence of noise, but even in very low signal-to-noise ratio (SNR = 0 dB), the uncertainty was not higher than $6~\mu \text{m}$ , which outperforms the COM estimate. The method can be implemented in image data as well as by using the raw signals. It is proposed that the signal-derived localization should replace the image-based equivalent, as it provides at least 10 times improved accuracy.