Ultrasound Shear Wave Elasticity Imaging With Spatio-Temporal Deep Learning

Abstract

Ultrasound shear wave elasticity imaging is a valuable tool for quantifying the elastic properties of tissue. Typically, the shear wave velocity is derived and mapped to an elasticity value, which neglects information such as the shape of the propagating shear wave or push sequence characteristics. We present 3D spatio-temporal CNNs for fast local elasticity estimation from ultrasound data. This approach is based on retrieving elastic properties from shear wave propagation within small local regions. A large training data set is acquired with a robot from homogeneous gelatin phantoms ranging from 17.42 kPa to 126.05 kPa with various push locations. The results show that our approach can estimate elastic properties on a pixelwise basis with a mean absolute error of 5.01 ± 4.37 kPa. Furthermore, we estimate local elasticity independent of the push location and can even perform accurate estimates inside the push region. For phantoms with embedded inclusions, we report a 53.93% lower MAE (7.50 kPa) and on the background of 85.24% (1.64 kPa) compared to a conventional shear wave method. Overall, our method offers fast local estimations of elastic properties with small spatio-temporal window sizes.

Ultrasound shear wave velocity estimation in a small field of view via spatio-temporal deep learning

“…For shear wave velocity estimation with deep learning, we consider a similar approach to Neidhardt et al 9 The authors propose to use a spatio-temporal CNN to estimate tissue elasticity directly from a sequence of 2D displacement images by performing a regression task between tissue elasticity or shear wave velocity and input image sequences. Hence, a function f : R h×w×t → R is learned, which maps an input image sequence x t to a shear wave velocity c ST-CNN , with w and h for the image width and height, respectively.…”

“…Hence, a function f : R h×w×t → R is learned, which maps an input image sequence x t to a shear wave velocity c ST-CNN , with w and h for the image width and height, respectively. We use a similar architecture as Neidhardt et al 9 and adapt a 3D version of the concept of Densely Connected Convolutional Networks (DenseNet). 16 We use three initial convolutional layers with five feature maps each.…”

“…Recently, deep learning methods have been proposed for US-SWEI. [6][7][8][9] It was demonstrated that shear wave velocity can be estimated from space-time respresentations 6 and that elasticity maps and segmentation masks can be generated from US-SWEI data. 7,9 In addition, first results on the estimation of viscoelastatic parameters using simulated shear wave motion data were shown.…”

“…[6][7][8][9] It was demonstrated that shear wave velocity can be estimated from space-time respresentations 6 and that elasticity maps and segmentation masks can be generated from US-SWEI data. 7,9 In addition, first results on the estimation of viscoelastatic parameters using simulated shear wave motion data were shown. 8 The advantage of these approaches is that data processing is performed in an end-to-end fashion and abstract features can be learned from the data.…”