Synthetic Elastography Using B-Mode Ultrasound Through a Deep Fully Convolutional Neural Network

Abstract

Shear-wave elastography (SWE) permits local estimation of tissue elasticity, an important imaging marker in biomedicine. This recently-developed, advanced technique assesses the speed of a laterally-travelling shear wave after an acoustic radiation force "push" to estimate local Young's moduli in an operator-independent fashion. In this work, we show how synthetic SWE (sSWE) images can be generated based on conventional B-mode imaging through deep learning. Using sideby-side-view B-mode/SWE images collected in 50 patients with prostate cancer, we show that sSWE images with a pixel-wise mean absolute error of 4.8 kPa with regard to the original SWE can be generated. Visualization of high-level feature levels through t-Distributed Stochastic Neighbor Embedding reveals a high degree of overlap between data from different scanners. Also qualitatively, sSWE results seem generalisable to single B-mode acquisitions and other scanners. In the future, we envision sSWE as a reliable elasticity-related tissue typing strategy that is solely based on B-mode ultrasound acquisition.

Deep learning in ultrasound elastography imaging: A review

“…This enabled the deep network to combine fine and course level information to generate high resolution synthetic SW elasticity maps, with less than 10% deviation in the clinically relevant elasticity range. 83 Figure 4 shows the structures of some main deep learning architectures.…”

“…It has evolved from the traditional CNNs and constitutes a general convolutional process on the contracting path and transposed 2D convolutional layers on the expansive path. Recently, Wildeboer et al 83 utilized this architecture to generate synthetic SWE maps from B-mode ultrasound images through end-toend nonlinear mapping. Their network contained direct skip connections on both the encoder filter layer and the decoder layer.…”

“…The model transferred the encoder layer output across the latent space and then concatenated the large‐scale features during decoding. This enabled the deep network to combine fine and course level information to generate high resolution synthetic SW elasticity maps, with less than 10% deviation in the clinically relevant elasticity range 83 . Figure 4 shows the structures of some main deep learning architectures.…”

Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency

“…By employing constrained CycleGAN, the experiment could also improve the accuracy of automatic segmentation using POCUS data. Wildeboer et al [ 53 ] presented methods of generating synthesized shear-wave elastography (SWE) images based on original B-mode images. Using both B-mode and SWE images collected from 50 prostate cancer patients, it was shown that synthesized SWE images with an average absolute error of pixel units within 4.5±0.96 kPa could be generated.…”

AUE-Net: Automated Generation of Ultrasound Elastography Using Generative Adversarial Network

“…Due to the excellent performance of GANs on natural image synthesis, GANs were introduced to perform various medical imaging tasks including solving problems associated with data imbalance of virtual STIR images [ 16 ], reducing metal artifacts and the radiation dose during digital tomosynthesis [ 17 ], as well as helping to process medical image [ 18 , 19 ]. Although neural networks and GANs were used to extract strain images from radio frequency data [ 20 , 21 ], and generate shear wave elastography images [ 22 ], we attempted to directly map conventional ultrasound images towards the corresponding strain elastography (SE) images. To the best of our knowledge, this is the first work to apply the nonphysical method to generate strain elastography images from conventional ultrasound images.…”