Rebholz Brandon, Almekkawy Mohamed
Annu Int Conf IEEE Eng Med Biol Soc. 2020 Jul;2020:2137-2141. doi: 10.1109/EMBC44109.2020.9175457.
Ultrasound images have an inherently low lateral resolution due to the size of transducers that are used in standard clinical scanners. This makes for low resolution images, as well as imprecise lateral displacement estimation. In speckle tracking, the well known discipline of estimating displacement by tracking pixel movement, lateral interpolation is often used to get subsample accurate displacement estimation. Standard methods for interpolation are known as inverse distance weighting methods, of which the well known cubic interpolation method is a part. Kriging interpolation, however, is a stochastic approach that uses statistical data to calculate interpolated data points as opposed to the purely mathematical methods of more traditional interpolators. This analysis tests the efficacy of one variety of Kriging interpolation, called Simple Kriging, on ultrasound data. Simple Kriging is tested on its accuracy to interpolate a sparse ultrasound image frame, as well as its usefulness in interpolating the correlation map to estimate subsample displacement. The applied bias of the estimation using Simple Kriging is also tested by interpolating the autocorrelation map where displacement is zero. Simple Kriging is an alternative interpolation scheme that could be used with image data and its accuracy is comparable to the accuracy of using the cubic interpolation.
由于标准临床扫描仪中使用的换能器尺寸，超声图像的横向分辨率天生较低。这导致图像分辨率低，以及横向位移估计不准确。在散斑跟踪中，这一通过跟踪像素移动来估计位移的著名方法，通常使用横向插值来获得亚采样精度的位移估计。标准的插值方法被称为反距离加权法，其中著名的三次插值法就是其中一种。然而，克里金插值是一种随机方法，它使用统计数据来计算插值数据点，这与更传统插值器的纯数学方法不同。本分析测试了一种名为简单克里金的克里金插值方法对超声数据的有效性。对简单克里金在插值稀疏超声图像帧方面的准确性，以及在插值相关图以估计亚采样位移方面的有用性进行了测试。还通过对位移为零的自相关图进行插值，测试了使用简单克里金进行估计时的应用偏差。简单克里金是一种可用于图像数据的替代插值方案，其准确性与使用三次插值的准确性相当。
