Training Variational Networks With Multidomain Simulations: Speed-of-Sound Image Reconstruction

  title={Training Variational Networks With Multidomain Simulations: Speed-of-Sound Image Reconstruction},
  author={M{\'e}lanie Bernhardt and Valery Vishnevskiy and Richard Rau and Orcun Goksel},
  journal={IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control},
Speed-of-sound (SoS) has been shown as a potential biomarker for breast cancer imaging, successfully differentiating malignant tumors from benign ones. SoS images can be reconstructed from time-of-flight measurements from ultrasound images acquired using conventional handheld ultrasound transducers. Variational networks (VNs) have recently been shown to be a potential learning-based approach for optimizing inverse problems in image reconstruction. Despite earlier promising results, these… 

Figures and Tables from this paper

Deep Learning for Ultrasound Speed-of-Sound Reconstruction: Impacts of Training Data Diversity on Stability and Robustness

It is demonstrated that training the network with a joint set of datasets including both geometrical and natural tissue models improves the stability of the predicted speed-of-sound values both on simulated and measured data.

DeepUCT: Complex cascaded deep learning network for improved ultrasound tomography

A deep learning model is developed using two cascaded convolutional neural networks with an encoder–decoder architecture to achieve near real-time image reconstruction in ultrasound tomography and significantly outperforms the state-of-the-art traditional iterative method both quantitatively and qualitatively.

Displacement Estimation Methods for Speed-of-Sound Imaging in Pulse-Echo

This work study and compare several conventional DE methods for the SoS reconstruction task with numerical and tissue-mimicking phantoms and evaluates SoS contrast dependency of reconstructions.

Ultrasound Signal Processing: From Models to Deep Learning

An overview of model-based deep learning techniques for medical ultrasound applications, which require less trainable parameters and training data than conventional neural networks, and concludes with a future perspective on these model-∗ based solutions.

Global Speed-of-Sound Prediction Using Transmission Geometry

—Most ultrasound (US) imaging techniques use spatially-constant speed-of-sound (SoS) values for beamforming. Having a discrepancy between the actual and used SoS value leads to aberration artifacts,

Estimating Mean Speed-of-Sound from Sequence-Dependent Geometric Disparities

In ultrasound beamforming, focusing time delays are typically computed with a spatially constant speed-of-sound (SoS) assumption. A mismatch between beamforming and true medium SoS then leads to

Anisotropic regularization of ultrasound pulse-echo tomography for reconstruction of speed-of-sound and tissue heterogeneity through abdominal layers

Regularization has a critical impact on the performance of pulse-echo computed-tomography and is a key component of the speed-of-sound (SoS) reconstruction pipeline. We investigate SoS reconstruction

Speed-of-sound imaging using diverging waves

Diverging waves (DW) is proposed for SoS imaging and its effects on the key step of displacement tracking in the SoS reconstruction pipeline, comparatively between PW and DW on a synthetic example, and the parameterization sensitivity of both approaches is presented on a set of simulated phantoms.

Nonlinear Waveform Inversion for Quantitative Ultrasound

—Due to its non-invasive and non-radiating nature, along with its low cost, ultrasound (US) imaging is widely used in medical applications. Typical B-mode US images have limited resolution and

Phase-Aberration Correction in Shear-Wave Elastography Imaging Using Local Speed-of-Sound Adaptive Beamforming

Shear wave elasticity imaging (SWEI) is a non-invasive imaging modality that provides tissue elasticity information by measuring the travelling speed of an induced shear-wave. It is commercially



Image Reconstruction via Variational Network for Real-Time Hand-Held Sound-Speed Imaging

A variational network based image reconstruction architecture that is based on optimization loop unrolling is presented that is able to reconstruct images with significantly different statistics compared to the training set, and an efficient training protocol of this network architecture on fully synthetic inclusion data is provided.

Deep Variational Networks with Exponential Weighting for Learning Computed Tomography

A network architecture that performs filtering jointly in both sinogram and spatial domains is introduced that is qualitatively and quantitatively superior to conventional non-linear reconstruction methods as well as state-of-the-art deep networks for image reconstruction.

Spatial domain reconstruction for imaging speed-of-sound with pulse-echo ultrasound: simulation and in vivo study

A SoS reconstruction strategy, where the forward problem is formulated using differential time-of-flight measurements based on apparent displacements along different ultrasound wave propagation paths, and the inverse problem is solved in spatial-domain using a proposed total-variation scheme with spatially-varying anisotropic weighting to compensate for geometric bias from limited angle imaging setup.

A Deep Learning Framework for Single-Sided Sound Speed Inversion in Medical Ultrasound

This paper presents an alternative approach using single-sided pressure-wave sound speed measurements from channel data using a fully convolutional deep neural network, and shows that it is possible to invert for longitudinal sound speed in soft tissue at high frame rates.

Speed-of-Sound Imaging Based on Reflector Delineation

A novel ultrasound computed tomography technique that can reconstruct local SoS in tissue on conventional ultrasound machines with hand-held linear arrays is developed and preliminary in-vivo results show the potential of this new technique to detect and differentiate malignant and benign lesions in the breast.

Deep ADMM-Net for Compressive Sensing MRI

Experiments on MRI image reconstruction under different sampling ratios in k-space demonstrate that the proposed novel ADMM-Net algorithm significantly improves the baseline ADMM algorithm and achieves high reconstruction accuracies with fast computational speed.

Image reconstruction by domain-transform manifold learning

A unified framework for image reconstruction—automated transform by manifold approximation (AUTOMAP)—which recasts image reconstruction as a data-driven supervised learning task that allows a mapping between the sensor and the image domain to emerge from an appropriate corpus of training data is presented.

Speed-of-sound imaging using diverging waves

Diverging waves (DW) is proposed for SoS imaging and its effects on the key step of displacement tracking in the SoS reconstruction pipeline, comparatively between PW and DW on a synthetic example, and the parameterization sensitivity of both approaches is presented on a set of simulated phantoms.

Ultrasound Aberration Correction based on Local Speed-of-Sound Map Estimation

This paper demonstrates a direct delay correction approach for US beamforming, by leveraging 2D spatial SoS distribution estimates from plane-wave imaging and proves that resolutions close to the wavelength limit can be achieved using the proposed local SoS-adaptive beamforming.

Improved forward model for quantitative pulse-echo speed-of-sound imaging.